Bone 37 (2005) 871 – 875 www.elsevier.com/locate/bone
Rapid Communication
Hip axis length as an independent risk factor for hip fracture independently of femural bone mineral density in Caucasian elderly Brazilian women Alberto Frisoli Jr. a,⁎ , Ana Patricia Paula b , Marcelo Pinheiro a , Vera Lucia Szejnfeld c , Ronaldo Delmonte Piovezan d , Ednilson Takata e , Tatiana Araújo Silva f , Paulo Henrique M. Chaves g a
Department of Medicine-Geriatric and Gerontology Division/Rheumatology Division-Federal University of São Paulo, Rua Pedro de Toledo 1010 São Paulo 04039-002, Brazil b Department of Clinical Medicine, Division of Rheumatology, Hospital Universitário de Brasília, University of Brazil, Brazil c Department of Medicine, Rheumatology Division, Federal University of São Paulo, Brazil d Department of Medicine, Geriatric and Gerontology Division, Federal University of São Paulo, Brazil e Department of Orthopedic, Federal University of São Paulo, Brazil f Department of Medicine, Geriatric and Gerontology Division, Federal University of São Paulo, Brazil g Departments of Medicine and Epidemiology, Johns Hopkins University, Center on Aging and Health, 2024 E. Monument Street, Suite 2-700, Baltimore, MD, USA Received 28 April 2005; revised 22 July 2005; accepted 9 August 2005 Available online 26 September 2005
Abstract Studies carried out in several countries and in different ethnic groups have suggested that the hip axis length (HAL) may be a risk factor for hip fractures. To evaluate if the HAL is an independent risk factor for hip fractures in elderly Caucasian Brazilian women, this study includes 112 participants sustaining proximal femur osteoporosis. Through HAL and bone mineral density (BMD) measurements, a statistical analysis using a multivaried regression curve was done. HAL was significantly longer in women sustaining a hip fracture than in the control group (99.24 ± 5.9 mm vs. 96.95 ± 5.6 mm, P b 0.05). After adjusting the standard HAL deviation for neck and trochanter BMD, OR was 1.43 (IC 95% 0.29–1.07; P b 0.08). When HAL was categorized for 97.8 mm (average HAL in all women), OR was 2.24 (IC 95% 1.04–4.84; P b 0.05). In conclusion, HAL may be associated with risk of hip fracture regardless of age, weight or BMD of elderly Brazilian Caucasian women. © 2005 Elsevier Inc. All rights reserved. Keywords: Elderly; Hip axis length; Hip fracture; Osteoporosis
Introduction The rate of hip fracture has increased as result of the aging in the population [1–3]. The number of people over 65 can reach one and a half billion in 2050 around the world [4], with an incidence of 6.26 million hip fractures per year [5]. Several studies have associated the increase in fracture risk with a decrease in BMD [6–8]. However, the predictive BMD value for fracture decreases in individuals over 70. De ⁎ Corresponding author. Fax: +5 11 5575 5710. E-mail address:
[email protected] (A. Frisoli). 8756-3282/$ - see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.bone.2005.08.004
Laet et al. [9] related that, from the age of 60 to the age of 80, the risk of hip fracture increases 13 times, while the decrease in BMD only justifies the risk rise only twice. Inter-racial and inter-ethnical studies also report significant variations in the rates of hip fracture regardless of BMD. Ross et al. [10] reported the number of hip fractures sustained by Japanese women from Hawaii or Japan as around 50% greater than in Caucasian Americans. A nationwide poll on women's hip fracture in the United States noticed that women of the Black race were 50% less vulnerable to fractures than Caucasians. The difference persisted even after the adjustment for age. Another phenomena described by Bauer [11] revealed that
872
A. Frisoli Jr. et al. / Bone 37 (2005) 871–875
non-Hispanic Caucasians displayed twice as big as hip fracture rate than Caucasian Hispanics. The BMD does not justify the variation found in the risk of hip fracture among the races and between the postmenopausal women and the elderly. For the elderly, some factors as propensity to falling, neurologic diseases, a decrease in sight acuity and muscular power, as well as balance disturbs have justified the disproportional increase in risk of fracture. The fracture differences among the races have been attributed to alimentary habits, variation in femoral geometry, propensity to falling and corporal composition [12,13]. The hip axis length (HAL), main component of femoral geometry, appears to be an indicator of the hip's capacity to absorb a fall's energy. The femoral momentum may or may not have any relation with the hip axis length as this depends on the circumstances of the fall (site and angle of impact) [14]. Several authors have evaluated the influence of HAL on the fracture rates of some races and ethnic groups and have found significant differences in the measurements of these women's HAL. In some studies, these variations were associated with the variation in the fracture rates [15–21]. In the present study, we investigated whether there is an association between HAL and risk of hip fracture in elderly Caucasian women living in Brazil.
objectives, as well as the likely inconveniences arising in the course of the study. The physician in charge of the study signed all consent forms. In women controls not sustaining a fracture, the scans were run at random, both in the right femur and in the left one. The positioning of the femur was made uniform for all subjects, by means of a hip positioning system (HPS) designed specially for the study at hand, in conformity with the device developed by Hans et al. [22]. The role of HPS was to keep the hip rotation inside stable in all women, reducing the coefficient of variance among the scans. The proximal femur BMD analyses was carried out by one sole observer, blind for fracture. The evaluation of the hip axis length was made in conformity with the technique described by Faulkner et al. in 1994 [23], which consists in determining the distance between the lower base of the greater trochanter, through the femur neck, as far as the inside brim of the iliac bone, done frontally (Fig. 1). It is carried out on the scan of the bone densitometry of the proximal femur through a computer program developed specifically for the densitometer DPXLUNAR, by the LUNAR® company. The precision error (reproducibility) of the hip axis length was determined with a group of 10 volunteer subjects, who carried out 3 successive scans, with the same device. UNIFESP Medical Ethics Committee approved the study.
Patients and methods Statistical analysis From January 1997 to June 1999, 162 women from Hip Diseases Nursery of the Orthopedics and Traumatology Department and in the Center of Osteometabolic Diseases of the Rheumatology Division of the Federal University of Sao Paulo, as well as nursing homes were interviewed. Caucasian Brazilian women aged 65 or older, who sustained no difficulty to walk, no bilateral hip prosthesis or any disease which might limit hip movement, such as rheumatoid arthritis, advanced osteoarthrosis, among others, were selected. Women with a previous diagnosis of any disease which might interfere in the metabolism of the bone or which might have affected the bone mineral density were left out. The inclusion criteria for osteoporosis was t score of BMD of the proximal femur lower or equal to −2.5 standard deviation. The hip fractures should not have been caused by trauma involving a high concentration of energy such as automobile accidents, falls from heights superior to that of the person or even caused by metabolic or neoplasic diseases in the past. Bone mineral density of proximal (BMD) femur was carried out of all women using dual-energy X-ray absorptiometry (LUNAR®, model DPX) scanner. The coefficient of variance for BMD of the femur neck was 2.9% and for the femur trochanter was 3.2%. Before running this BMD analysis, each subject's weight (kg) and height (cm) were obtained using an antroprometric scale tuned daily. The women selected for the study signed consent forms, which contained explanations of the procedures and their
The age, weight, height, BMD of the neck and of the trochanter and HAL of the women in both groups were
Fig. 1. Hip image in frontal plane with hip axis length and femur axis length.
A. Frisoli Jr. et al. / Bone 37 (2005) 871–875
expressed in average and standard deviation terms (X ± SD). The inferential analysis of both populations was carried out through Student's t test. The chosen significance level was P b 0.05 (alpha = 5%). Odds ratio (OR) was calculated from a multivaried logistic regression analysis, used to describe the association between the measurements of independent variables (bone mineral density of the neck and trochanter and the hip axis length) and hip fracture occurrence. (If the interval of reliability of 95% for the odds ratio did not include 1.0, the analyzed ratio was considered as being associated with hip fracture.) The multivaried logistic models have been used to adjust possible confounding factors, such as age and height. The correlation existing between the continuous variables was carried out, using Pearson's variance coefficient. The significant correlation coefficient was considered only when greater than or equal to 0.40 (P b 0.05). The statistical analyses were made using software named Statistica (version 5, 1997).
Results The sample was made up of 112 women with the diagnosis of osteoporosis, of which 46 sustained a hip fracture (31 neck and 15 transtrochanteric) and 66 controls did not sustain a fracture. The characteristics of both women sustaining a fracture or not are listed in Table 1. There were no statistically significant differences between the ages, weights and heights of both groups. There were statistically significant differences between the average BMD of the femur neck (BMD hip fracture = 0.612 ± 0.09 vs. BMD control = 0.624 ± 0.05 g/cm2; P = 0.041) of both women, sustaining and not sustaining a fracture. The same was not seen in the femur trochanter (BMD hip fracture = 0.563 ± 0.08 vs. BMD control = 0.544 ± 0.06 g/cm2; P = 0.20). The average HAL of those women sustaining a fracture was significantly higher than that of those women not sustaining one (HAL hip fracture = 99.24 ± 5.9 vs. HAL control = 96.95 ± 5.6 mm; P b 0.05). The precision error (reproducibility) of the measurements of HAL was 1.8 ± 0.5 mm, namely 1.84%.
Table 1 The characteristics of women sustaining and not a hip fracture (X ± SD) Variables
No hip fracture (n = 66)
Hip fracture (n = 46)
P
Age (years) Height (cm) Weight (kg) BMD of neck (g/cm2) BMD of trochanter (g/cm2) HAL (mm)
77.35 ± 6.79 147.48 ± 6.44 57.53 ± 10.94 0.62 ± 0.05
76.3 ± 7.66 147.89 ± 9.02 56.50 ± 12.89 0.61 ± 0.09
0.43 0.78 0.65 0.41
0.54 ± 0.06
0.56 ± 0.08
0.20
99.24 ± 5.94
0.04
96.95 ± 5.6
873
For carrying the correlation tests and the logistic regression analysis out, the variables BMD of the neck and trochanter were transformed into isolated variables and categorized for the cut-off points according to the average of the total sample's distribution. After such transformation, these variables did not display any statistical significance. For this reason, they were used in a continuous fashion, with a need for multiplying the BMD of femur neck and trochanter by 10, so as to increase the sample's stability. As such, these variables will appear multiplied by 10, as neck BMD (×10) and trochanter BMD (×10). Correlation tests among variables such as weight, height, age, hip axis length, neck BMD (×10) and trochanter BMD (×10) were carried out using Pearson's correlation coefficient, in the total sample, in the population of both women sustaining or not a fracture. In the population of women not sustaining a fracture, there was a statistically significant correlation between their HAL and height (r = 0.45; P b 0.0001). For the initial model of the multivaried logistic regression analysis, all variables were listed for the hip fracture event. In the final model, the only variables that remained stable were the HAL, the BMD of neck and trochanter (×10). HAL was categorized for the standard deviation (HALSD) of the sample, namely, 5.827 mm. The categorizing for the HAL standard deviation of the total sample was aimed at decreasing the assessment error of HAL as it is carried out over a two-dimensional image, without precise anatomical parameters, in which the possibilities of millimetric errors are great. Hip axis length odds ratio categorized for a standard deviation (5.827 mm), adjusted for continuous BMD of the neck and the trochanter, was 1.43 (IC 95% 0.99–2.06; P = 0.05), namely, given that a woman sustains hip fracture, her chances of displaying a 5.82 mm increase in the hip axis length are 1.43 greater. When femur BMD (×10) is analyzed against (adjusted) for HAL categorized for a standard deviation, neck BMD (×10) displays 0.57 OR (IC 95% 0.29–1.10; P = 0.09), and trochanter BMD (×10) OR was 1.92 (IC 95% 1.01–3.67; P = 0.04). In an additional analysis, when HAL was categorized for the average of HALs (97.8 mm) of the total sample, without an adjustment for the BMD of the proximal femur, it was noticed that OR was 2.24 (IC95% 1.04–4.84; P b 0.04).
Discussion In our sample, average HAL of women sustaining a fracture was significantly higher than of those not sustaining one, confirming data reported by other authors, who have also encountered statistically significant difference between the average HAL of the groups sustaining or not a fracture and using a similar technique to ours [24].
874
A. Frisoli Jr. et al. / Bone 37 (2005) 871–875
When we consider the average of the HALs of the total sample (97.8 mm), we observe similar results to those of Chin et al. [17] in Caucasian women residing in New Zealand, not sustaining a hip fracture, whose average of the HAL was 102.3 ± 5.3 mm, using the same measurement technique as that employed by us. Faulkner et al. in 1993 [14], in a prospective study with 134 non-Black elderly women, by means of hand measuring the printed scan of the femur's bone density examination, also find HAL averages which were significantly different among those women sustaining a fracture and those not (67.0 ± 3.8; versus 69.3 ± 4.0 mm, respectively). The difference between these readings and the ones discovered by us and Peacock et al. [24] results from the differences in the techniques used, as Faulkner et al. measured HAL in the hip scan. Later, Faulkner et al. automatically evaluated HAL in the proximal femur bone densitometry of 198 women not sustaining a hip fracture, and the average of HAL of American women between 40 and 92 was 105 mm. In our study, unlike the others, all women presented osteoporosis in the femur neck, which underscores the relevance of HAL. For Peacock et al. [24], in the multivaried regression analysis, OR for every standard deviation (6.8 mm) of HAL, after adjusting for BMD of neck and femur, was 1.85 (IC95% 1.02–3.45; P b 0.05). The difference between our data and the one published in 1993 by Faulkner et al. [14] probably results from the heterogeneity of the samples studied by them, from which only Black women were excluded. The techniques were also different. On the other hand, the discrepancy between our data and that of Peacock et al. may be a consequence to the small size of the sample studied by these authors, 22 Caucasian women sustaining a fracture and 43 not sustaining one. Until our study, he was the only author who studied Caucasian women only. The genetic contribution is possibly the most important determinant of the HAL. However, eating habits and lifestyle may also contribute to the diversity found among our data, that of Chin et al. [17] and that of Peacock et al. [24]. The rising globalization of cultures along the years is likely to have softened the differences in lifestyles among the various peoples, thus affording a smaller variation in the obtained HALs. This, as a matter of fact, seems to be underway. Other aspects, which ought to be emphasized, the average height and weight of our samples, were not significantly different from those of other studies with elderly Caucasian North American and European women [25]. Several authors, by means of studies with women from different races, have come across significant variations in the measurement of HALs. Asian women present shorter HAL than those in Black women, which in turn, present shorter HAL than those in Caucasian. The shorter HAL in the Asian, when compared to those in Caucasian, has been considered as a protecting factor in regards to the hip fracture risk. This was widely demonstrated by both Cummings et al. [15] and Nakamura et al. [19]. These authors verified that Asian
women, although with lower BMD than Caucasian, presented fracture risk 47% lower (IC 95%; 32–63%). Daniels et al. [16] have found lower proximal femur BMD and shorter HAL in Black women than in Caucasian. However, Cummings et al. have reported that the hip fracture risk in Black women was 32% lower than in Caucasian, although it was higher than in Asian whose BMD is lower. These authors have concluded that a shorter HAL would act as one of the main factors of the high fracture risk encountered in Black women, regardless of BMD. The great variation of races and ethnicities present in Brazil, as well as its mix, may have given origin to a population which, however, phenotypically Caucasian, possesses genetic influences from the Black and the indigenous. This has generated a population of Caucasian women with characteristics that distinguish them from the populations studied in other countries. Such fact may explain the difference between the HALs we have committed to studying and the ones reported by other authors. When we use Pearson's correlation test, we were able to observe that the only positive and statistically significant correlation was that between the HAL and height of women not sustaining a fracture, which confirms the findings of other authors [14,15,17]. This correlation was not observed among women sustaining a fracture probably because these women presented a more significant loss of height due to sequels from the fracture itself. This has been described in women as a means of reducing hip by up to 3 cm high due to surgical corrections and hip repositioning. Different from the data found in other studies, trochanter BMD observed in patients sustaining a hip fracture presented a positive correlation for the latter, when adjusted for BMD of neck. This diversity was attributed to the heterogeneity and size of the sample of women sustaining a fracture. This study presents some restrictions. The formed sample may not correspond to all women who sustain a hip fracture as a result of osteoporosis. As an inclusion criteria, the women were neither to possess a bilateral hip prostheses nor any important limitation for strolling. This ultimately restricts sampling by at least 78% of the universe of women sustaining a fracture as 22% of them become unable to stroll following hip fracture [26]. Another impairment in this sample is the high mortality rate in patients in the first 6 months of fracture, namely, 16%. One piece of criticism in regards to the homogeneity of the sample results from not questioning the ladies on the possibility of race mixing in future generations. With regard to the type of study, the case control evaluates the possibility of a variable being present in an event, although the best way of quantifying the predictive risk of a variable to an event would be by means of carrying out a prospective study. The HAL evaluation has a number of advantages. It may be measured in proximal femur bone densitometry scans, without causing an increase in costs or time for carrying out the scan. Besides, HAL may increase predictive BMD in specific sites for hip fracture. As we view, its use isolatedly
A. Frisoli Jr. et al. / Bone 37 (2005) 871–875
would be interesting as an identifier of risk population, while, in the adjusted form for BMD, it may be useful for stratifying the risk of elderly individuals to falling. Prospective studies are still needed, including men as well as women from other races. These further data could evaluate the importance of HAL isolatedly in determining hip fracture in our surroundings and whether the association of the risk established by BMD with the risk of HAL may be improved in predicting hip fracture, but not in the vertebral fracture, as demonstrated in 1995, by Boonen and collaborators [27]. We conclude that HAL of elderly Brazilian Caucasian women sustaining a fracture is significantly longer than of those not sustaining a fracture. The HAL presented a positive and statistically significant correlation with hip fracture when analyzed isolatedly and when adjusted for neck BMD.
[13]
[14]
[15]
[16]
[17]
[18]
References [19] [1] Lancaster HO. Expectations of Life. Berlin: Springer; 1990. [2] Cummings SR, Black DM, Nevitt MC. Bone density at various sites for prediction of hip fractures. Lancet 1993;341:72–5. [3] Marshal D, Schnell O, Wedel H. Meta-analysis of how well measures of bone mineral density predict occurrence of osteoporotic fractures. BMJ 1996;312:1254–9. [4] Cooper C, Campion G, Melton LJ. Hip fractures in the elderly: a worldwide projection. Osteoporosis Int 1992;2:285–91. [5] Gulberg B, Johvell O, Kanis JA. Worldwide projections for hip fracture. Osteoporosis Int 1997;7:407-4136. [6] Melton III LJ, Wahner HW, Richelson LS, O'Fallen WM, Riggs BL. Osteoporosis and the risk of hip fracture. Am J Epidemiol 1986;124: 254–61. [7] Melton III LJ, O'Fallen WM, Riggs BL. Secular trends in the incidence of hip fractures. Calcif Tissue Int 1987;41:57–64. [8] Ross PD, Davis JW, Vogel JM, Wasnich R. A critical review of bone mass and the risk of fracture in osteoporosis. Calcif Tissue Int 1990; 46:149–61. [9] De Laet CE, Van Hant BA, Burger H. Bone density and risk of hip fracture in men and women: cross sectional analysis. BMJ 1997; 315:221–5. [10] Ross PD, Norimatsu H, Davis JW, Yano K, Wasnich RD, Fujiwara S, et al. A comparison of hip fracture incidence among native Japanese, Japanese Americans, and American Caucasians. Am J Epidemiol 1991;133(8):801–9. [11] Bauer RL. Ethnic differences in hip fracture: a reduced incidence in Mexicans Americans. Am J Epidemiol 1988;127(1):1482–3. [12] Orimo H, Hosoi T, Nakamura T, Ouchi Y, Shirakil M. Epidemiology of fractures in Asia. In: Christiansen C, Overgaard K, editors.
[20]
[21]
[22]
[23]
[24]
[25] [26]
[27]
875
Osteoporosis; 1990. p. 55–61. [Handelstrykkeriet Aalborg Aps, Alborg, Dinamarca]. Beck TJ, Ruff CB, Warden KE, Scott WW, Rao G. Predicting femoral neck strength from bone mineral data: a structural approach. Invest Radiol 1990;25:6-18. Faulkner KG, Cummings SR, Black DM, Palermo L, Gluer CC, Genant HK. Simple measurement of femoral geometry predicts hip fracture: the study of osteoporotic fractures. J Bone Miner Res 1993;8 (10):1211–7. Cummings SR, Cauley JA, Palermo L, Ross PD, Wasnich RD, Black DM, et al. For the study of osteoporotic fractures research group. Racial differences in hip axis lengths might explain racial differences in rates of hip fracture. Osteoporosis Int 1994;4:226–9. Daniels ED, Dehifor JM, Schintzler CH, Moodly GP, Zochen D. Differences in mineral homeostasis, volumetric bone mass and femoral neck axis length in Black and White South African women. Osteoporosis Int 1997;7:105–12. Chin K, Evans MC, Cornishit R, Cundy T, Reid IR. Differences in hip axis and femoral neck length in premenopausal women of Polynesian, Asian and European origin. Osteoporosis Int 1997;7:344–7. Mikhail MB, Vaswan NA, Aloie JF. Racial differences in femoral dimensions and their relation to hip fracture. Osteoporosis Int 1996;6:22–4. Nakamura T, Turne CH, Yoshikawa T, Slemenda CW, Peachock M, Burr DB, et al. Do variations in hip geometry explain differences in hip fracture risk between Japanese and White Americans? J Bone Miner Res 1994;9(7):1071–5. Villa ML, Marcus R, Delay RR, Kesley JL. Factors contributing to skeletal health of postmenopausal Mexican–American women. J Bone Miner Res 1995;10(8):1233–42. Wang MC, Aguirre M, Bhudhkasok GS, Kendall CG, Kirsh S, Marcus R, et al. Bone mass and hip axis length in healthy Asian, Black, Hispanic, and White American youths. J Bone Miner Res 1997;12 (11):1922–30. Hans D, Doboeuf F, Schott AM, Avioli LV, Drezner MK, Meunier PJ. Effects of a new positioner on the precision of hip bone mineral density measurements. J Bone Miner Res 1997;12(8):1289–94. Faulkner KG, McClung M, Cummings SR. Automated evaluation of hip axis length for predicting hip fracture. J Bone Miner Res 1994;9 (7):1065–9. Peacock M, Turner CH, Liu G, Manatunga AK, Timmerman L. Better discrimination of hip fracture using bone density, geometry and architecture. Osteoporosis Int 1995;5:167–73. Faulkner KG. Hip axis length and osteoporosis fractures—Letter to editor. J Bone Miner Res 1995;10(3):506–8. Greendale GA, Barret-Connor E, Ingles S, Haile R. Late physical and functional effects of osteoporotic fracture in women: the Rancho Bernardo Study. J Am Geriatric Soc 1995;43:955–61. Boonen S, Koutri R, Dequeker J, Aerssens J, Lowet G, Nijs J, et al. Measurement of femoral geometry in type I and type II osteoporosis: differences in hip axis length consistent with heterogeneity in the pathogenesis of osteoporotic fractures. J Bone Miner Res 1995;10 (12):1908–12.
