SEARCH

SEARCH BY CITATION

Keywords:

  • aged;
  • cohort study;
  • risk assessment;
  • fractures;
  • primary care

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The aim of this prospective study was to develop a risk score, based on putative risk factors in current guidelines, which can be used to identify women at high risk of fractures in general practice. The study sample included 4157 women ≥60 yr of age (mean ± SD: 74.1 ± 9.1 yr), with a median follow-up of 8.9 yr of the Rotterdam Study (ERGO), and 762 women ≥65 yr of age (mean ± SD: 76.0 ± 6.7.yr), with a median follow-up of 6.0 yr of the Longitudinal Aging Study Amsterdam (LASA). Potential risk factors were those proposed in risk scores of three recent guidelines on osteoporosis: age, family history of fractures, prior fracture, low body weight/body mass index (BMI), serious immobility, rheumatoid arthritis, current smoking, alcohol consumption >2 units daily, prevalent vertebral fracture, and systemic corticosteroid use. Five-year absolute risk of hip fracture was 3.9% in the Rotterdam Study and 3.1% in LASA, and 10-yr absolute risk of hip fracture was 8.4% in the Rotterdam Study. Using Cox regression analysis, age (70–79 and 80+ versus <60–69) and four other risk factors were included in the risk profiles of hip fractures and fragility fractures: any prior fracture after age 50, body weight <64 kg, use of a walking aid as a proxy measure of serious immobility, and current smoking. Estimated 10-yr absolute risk of hip fracture ranged from 1.4% in women, age 60–69 years, without any of these predictors to 29% in women, ≥80 yr of age, having two or more positive risk factors. A simple risk score can satisfactorily identify older women at high risk of osteoporotic fractures in general practice. Future studies are needed to validate this score.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Osteoporosis is a major public health problem; worldwide, the number of fractures is increasing because of the rapidly increasing number of older people.(1) About 40% of women and 13% of men ≥50 yr of age will experience at least one fracture during their remaining lifetime, based on average life expectancy.(2,3) Because fractures are associated with impaired quality of life, physical decline, mortality, and high costs,(4–6) prevention of fractures is desirable. Effective interventions for the prevention of fractures are now available including bisphosphonates and selective estrogen receptor modulators (SERMs).(7,8)

Recently national(9,10) and international (WHO) guidelines(11,12) on the management of osteoporosis have been developed that recommend a “case-finding” approach to identify persons with a high risk of osteoporotic fractures in general practice. In the guidelines, various risk assessment scores have been developed based on risk factors, identified from literature review and meta-analyses.(13–17) Use of risk assessment scores can guide an appropriate use of BMD testing needed to prior intervention.(9–12) Despite overlap across the different guidelines, risk factors differ in each guideline and none of the proposed risk scores has yet been calibrated for use in general practice. To be used for “case-finding” in general practice, risk factors must be easily measurable and sufficiently prevalent or strongly related to fracture risk. In addition, the risk score must be as simple as possible, preferably simpler than the risk score of the WHO guideline (FRAX score).

The aim of this study was to develop a fracture risk assessment tool for use in general practice based on risk factors proposed in recent guidelines.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Study samples

We used two prospective cohort studies: the Rotterdam Study and the Longitudinal Aging Study Amsterdam (LASA).

The Rotterdam Study is a prospective, ongoing population-based cohort study of men and women ≥55 yr of age residing in the Ommoord district of the city of Rotterdam, in The Netherlands. The study was designed to investigate chronic, disabling diseases. The rationale and design of the study have been described previously.(18) The baseline examination included 7983 subjects (response rate 78%). This study was performed with 4157 female respondents ≥60 yr of age who participated in the baseline examination (between 1991 and 1993).

The LASA is an ongoing cohort study on predictors and consequences of changes in autonomy and well being in older men and women in the Netherlands. The sampling and data collection procedures have been described in detail elsewhere.(19,20) Briefly, a sample of older individuals (55–85 yr of age), stratified by age, sex, and urbanization, was drawn from the population registers of 11 municipalities in areas in the west, northeast, and south of the Netherlands. Every 3 yr, a LASA examination was performed, including a main and medical interview. The baseline examination (1992 or 1993) included 3107 individuals (response rate: 81.7%). This study was performed within a subsample of 779 female respondents who participated in the medical interview of the second data collection cycle (1995 or 1996) and who were ≥65 yr of age as of January 1, 1996. Seventeen of these 779 respondents were excluded because of missing data on both hip and fragility fractures.

All respondents in the Rotterdam and LASA Studies gave written informed consent, and the Medical Ethics Committee of the Erasmus Medical Center and of the VU University Medical Center approved these studies.

Assessment of fractures

In the Rotterdam Study, the occurrence of incident fractures was continuously monitored through general practitioners by means of a computerized system. Events were classified independently by two research physicians according to the International Statistical Classification of Diseases and Related Health Problems (ICD-10; International Statistical Classification of Diseases 1992). Finally, a medical expert in the field reviewed all coded events for a final classification. Information on incident nonvertebral fractures was collected from baseline (1990–1993) until January 1, 2002.

In LASA, fractures that occurred between the second examination (1995/96) and the third examination (1998/99) were reported every 3 mo prospectively with a fall and fracture calendar.(21) Information about fractures was obtained retrospectively from respondents who did not participate in the calendar follow-up. Fractures that occurred between the third examination and the fourth data examination (2001/02) were retrospectively assessed in 2001/02. For this study, assessment of fractures ended September 1, 2002. Ninety percent of all reported fractures were confirmed by the general practitioner. When a participant died or was lost to follow-up, the general practitioner was asked whether a fracture had occurred since the last contact with the respondent.(22)

We used hip fracture and fragility fracture, including hip, pelvis, proximal humerus, and wrist fractures, as outcomes. Fractures that were considered to be non-osteoporotic (i.e., fractures of the hand, foot, and head, fractures caused by a motor vehicle accident, and fractures caused by cancer [only in the Rotterdam Study]) were excluded. The duration of follow-up was recorded for each respondent from the date of enrollment in the study to the date of the first fracture, the date of death, the date of last contact with the respondent, or the date of the last follow-up.

Risk scores from national and international guidelines

The risk score of the Dutch Institute for Health Care (CBO)(9) includes fracture since age 50, prevalent vertebral fracture, low body weight (<60 kg), serious immobility, and corticosteroid use (≥7.5 mg prednisone equivalent/d). The CBO score is based on the sum of the score for each of the variables included in the profile. The risk score of the Dutch Standard of General Practitioners (NHG risk score)(10) includes age, fracture since age 50, prevalent vertebral fracture, positive family history, low body weight (<60 kg), serious immobility, and systemic corticosteroid use (≥7.5 mg prednisone equivalent/d). Risk factors used in the FRAX risk score of the WHO(11,12) were age, sex, low body mass index (BMI), prior fracture, a parental history of hip fracture, current smoking, ever use of systemic corticosteroids, high alcohol intake (>2 units/d), and the presence of rheumatoid arthritis (RA) (Table 1).

Table Table 1.. Summary of the Risk Factors in Three Recent Guidelines on Osteoporosis(9–11)
Thumbnail image of

Data collection for risk factors

For the Rotterdam Study, an extensive home interview was performed by trained interviewers between 1990 and 1993 in which questions were asked on history of fracture in the past 5 yr, prevalent vertebral fracture, family history of hip fractures, serious immobility, smoking, and alcohol intake. Data on systemic corticosteroid use and the presence of rheumatoid arthritis were obtained from the general practitioners (GPs). During a clinical examination at the research center, body weight was measured with respondents wearing indoor clothing and without shoes. BMI was computed as weight in kilograms divided by height in meters squared (kg/m2).

For LASA, trained interviewers or research nurses visited the respondents at home for a main and a medical interview between 1995 and 1996. Questions were asked on history of fracture since age 50, prevalent vertebral fracture, family history of hip fractures, serious immobility, smoking, and alcohol intake. The presence of joint complaints (including RA, osteoarthritis, and other) was assessed by self-report.(23) Use of systemic corticosteroids was assessed by recording the medications of the respondent directly from the containers. Body weight and BMI were measured similarly as in Rotterdam.

In both studies, operationalizations of the risk factors age, family history, low body weight/BMI, high alcohol intake, and smoking were done in accordance with the national and WHO FRAX risk scores (Table 1). Family history was defined as hip fracture of first family member in line with the Dutch NHG guideline; this operationalization predicted fragility fractures slightly better than the WHO definition, which is hip fractures of parents. For low body weight, we also included body weight <64 kg, because this seemed to be the best cut-off point for the prediction of fragility fractures.(24) In both the Rotterdam and LASA studies, smoking was defined as current smoking. Prevalent vertebral fracture was defined as self-reported vertebral fracture. We decided to use the self-reported measure to allow use of the risk score in general practice, where lateral spinal radiographs are not available for most patients. In the Dutch risk profiles, prior fracture and serious immobility were not further operationalized. Therefore, we used the most practical, easily measurable operationalizations.(24) In the Rotterdam Study, prior fracture was defined as any prior fracture in the past 5 yr, and serious immobility was defined as use of a walking aid. In the LASA Study, prior fracture was defined as any prior fracture since age 50 and serious immobility was defined as use of a walking aid during a walking performance test. In both the Rotterdam and LASA studies, corticosteroid use was defined as use of systemic corticosteroids at baseline.

Statistical analysis

To increase statistical power, data from the Rotterdam and LASA study were merged. Both studies included similar baseline data. Kaplan-Meier curves were used to examine 5- and 10-yr absolute risk of hip and fragility fracture for each potential risk factor. Risk factors of hip and fragility fractures were studied using Cox proportional hazard regression models. First, univariable HRs were calculated with 95% CIs. HRs can be interpreted as RRs. Risk factors were combined into a multivariable Cox regression model. The full model was simplified according to statistical strength (exclusion if p < 0.20), correlations between risk factors, and practical considerations. To facilitate clinical interpretation, the regression coefficients were transformed (multiplied by 2 and rounded of to the nearest integer) into simple scores that can be added up to obtain a sum score. Discriminative ability was expressed with a concordance (c) statistic, which is similar to the area under the receiver operating characteristic curve when dichotomous outcomes are considered. The c statistic can range from 0.5 to 1.0 for sensible models.(25) The regression model was performed on all included cases after missing values had been imputed with multiple imputation techniques.(26,27) We imputed five sets of data that were identical with regard to known information but could differ on imputed values for missing information. The imputation models were used for all the variables that we considered potential risk factors as well as for the occurrence of fractures during follow-up. Models were constructed in each of the five completed data sets and internally validated using bootstrapping techniques. The average of the bootstrap validated c-index was reported. Results were compared with analyses with complete cases and no relevant differences were found. The c-statistic of the risk score was compared with that of the WHO FRAX score. Analyses were performed with the Statistical Package for the Social Sciences (version 11.0; SPSS, Chicago, IL, USA) and R software (R 2.5.1).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The Rotterdam sample included 4157 women with a mean age of 74.1 yr. The LASA sample included 762 women with a mean age of 75.8 yr (Table 2). The distribution of most risk factor variables was similar between both study samples, except for the risk factors that were measured differently (prior fracture, use of a walking aid, and presence of RA).

Table Table 2.. Baseline Characteristics of Study Subjects (Women ≥60 yr of Age)
Thumbnail image of

In the Rotterdam Study, hip fractures occurred in 272 respondents, and fragility fractures occurred in 399 respondents during 31,472 person-years of follow-up (median, 8.9 yr; interquartile range [IQR], 5.5–10 yr). In the LASA study, 21 and 52 respondents, respectively, suffered a hip or fragility fracture during 3935 person-years of follow-up (median, 6.0 yr; IQR: 5.0–6.1 yr). Five-year absolute risk of hip fracture was 3.9% in the Rotterdam Study and 3.1% in LASA, and 10-yr absolute risk of hip fracture was 8.4% in the Rotterdam Study.

Table 3 shows the univariable HRs and 5- and 10-yr Kaplan-Meier risks of hip and fragility fractures for the risk factors that were examined. For most of the considered risk factors, except for current smoking, alcohol intake, and RA, there was a significant association with both hip and fragility fractures, because the 95% CIs did not include 1.

Table Table 3.. Contribution of Potential Risk Factors of Hip and Fragility Fractures for Respondents of the Rotterdam Study and LASA Study (n = 4919)
Thumbnail image of

All variables, except self-reported vertebral deformities and corticosteroid use, were entered into a multivariable regression model. Self-reported vertebral deformities and systemic oral corticosteroid use were too infrequent to perform statistical analysis. Low body weight and low BMI were strongly correlated. Preference was given to low body weight, because it predicted fragility fracture as well as low BMI,(23) and it was considered easier to measure. Subsequently, in both the models of hip fracture and fragility fracture, family history of hip fracture, drinking >2 units of alcohol daily, and presence of RA were excluded from the models, because of their limited predictive value (p > 0.20).

The variables that were included in the risk score of hip fracture were age (60–69, 70–79, 80+ yr), prior fracture, body weight <60 kg, use of a walking aid, and current smoking (Table 4). The same variables were included in the risk score of fragility fracture (Table 4). Low body weight had slightly more predictive value at a cut-off point of 64 kg (<64 versus ≥64 kg). Any prior fracture, low body weight, use of a walking aid, and current smoking had the same scores (score = 1). The sum score for hip fractures ranged from 0, when none of the predictors were present, to 8 when all predictors were present. The sum score for fragility fracture ranged from 0 to 7. The observed c-statistic for the risk scores of hip fractures was 0.77 both for using the regression coefficients and the transformed score, suggesting high discriminative ability to use these models in practice. The risk score predicted fragility fractures to a modest extent with an observed c-statistic of 0.71 using the regression coefficients and 0.70 using the transformed score. Bootstrapping was performed and a negligible optimism was shown. FRAX led to a similar c-statistic as a more selective model that omitted family history of fracture, daily alcohol use, and presence of RA (c = 0.76 for hip fractures).

Table Table 4.. Multivariable Cox Regression Analysis of Hip Fracture and Fragility Fracture in n = 4919 Respondents of the Rotterdam Study and LASA Study
Thumbnail image of

We made risk score tables and a risk score program to calculate 10-yr absolute risk of hip and fragility fracture among different levels of the risk score to enable health care professionals to easily use the risk score to identify women with high risk (Tables 5 and 6). Because both the national guidelines and WHO guideline recommend to treat women who use oral corticosteroids in a moderate dose with a bisphosphonate, and several previous studies have identified it as a very strong risk factor, we decided to add this variable to the final risk score. Because age weighted strongly, the calculation of the sum risk score was stratified by age. Absolute risks of fracture were from 1.4% for hip fracture and 5% for fragility fracture in women 60–69 yr of age without any of the four risk factors. The risks were 29% and 35% in women, ≥80 yr of age, having two or more of the four risk factors. We decided to define women 60–79 yr old with two or more risk factors and women ≥80 yr of age with one risk factor as having a high 10-yr absolute risk, prompting DXA scanning.

Table Table 5.. Ten-Year Absolute Risk of Hip Fracture in Women, ≥60 yr of Age Among Different Levels of the Risk Score
Thumbnail image of
Table Table 6.. Ten-Year Absolute Risk of Fragility Fracture in Women, ≥60 yr of Age Among Different Levels of the Risk Score
Thumbnail image of

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

This study provides a simple risk score for the assessment of hip and/or fragility fractures in primary care. The risk score is based on risk factors that are proposed in recent guidelines for treatment on osteoporosis(9–12) and included age (70–79 and 80+ versus <60–69), prior fracture after age 50, body weight <64 kg, use of a walking aid, and current smoking. Ten-year absolute risk of hip fracture ranged from 1.4% in women, age 60–69 yr, without any of these risk factors, to 29% in women, ≥80 yr of age, having two or more predictors.

This study has several strong points. First, we merged data of two longitudinal studies with up to 10- and 6-yr fracture follow-up in which comparable questionnaires were used for data collection. The large sample permitted examination of the predictive performance for most potential risk factors of recent risk scores with sufficient power. Because of the long follow-up, we were able to calculate 5- and 10-yr absolute risks of hip and fragility fractures. Fracture risk has traditionally been expressed as relative risks, but recently it has been recommended to express risk as 10-yr absolute risk.(9–12,28–30)

The risk score included four of the eight risk factors that were included in the WHO FRAX risk score(11,12): age, prior fracture, smoking, and systemic corticosteroid use. Other risk factors were low body weight and immobility, which were included in the risk scores of the two national guidelines. Some of the risk factors, including positive family history, alcohol use, and RA, that were included in the FRAX risk score were not included in the GP risk score, because they were not or hardly associated with risk of hip or fragility fractures and/or because they were difficult to assess in general practice. In line with results of a large meta-analysis,(14) a parental history of fracture was only weakly associated with risk of hip and fragility fracture. This may be explained by the unreliability of the measurement of this risk factor and the potential bias of longevity of the parents on the chance to have a fragility fracture.(24) Although it is well known that patients with RA have an increased risk of fractures,(31) no association was found in this study. This may also be explained by the inaccuracy of this measure when it is assessed in general practice or by self-report.(23,32) In general practice, RA is often under-reported, whereas over-reporting occurs by self-report, because older persons often do not know whether they have RA, osteoarthritis, or arthralgia. In contrast to a meta-analysis,(17) but in line with some other previous studies,(33,34) alcohol intake was not related to fracture risk. In general practice, high alcohol intake is difficult to measure because of social desirable answers.

Both national guidelines include vertebral fractures as a risk factor for fractures, because in a previous study using data of the Rotterdam Study, radiographically detected vertebral fractures were associated with a high risk for nonvertebral fractures.(35) However, we did not include vertebral fracture in this risk score, because in general practice, radiographs are not available for most patients. Self-reported vertebral fractures were not included because of their low prevalence and the inaccuracy and difficulty to measure it in general practice.

Our results have practical implications for “case-finding” osteoporosis in general practice or other primary care facilities. A general practitioner or nurse practitioner can easily use the fracture risk score. The risk factors can be quickly assessed by asking five simple questions. The 10-yr absolute risk can be determined from a risk table or with aid of a computer program (http://survey.erasmusmc.nl/osteoporose). Women with a high risk of hip and/or fragility fracture can be referred for BMD testing and/or can be recommended to take drugs, including bisphosphonates and/or calcium and vitamin D supplementation.

The c-statistic, which is similar to the area under the ROC curve, was 0.77 for hip fracture, indicating high discriminative ability to use this model in practice. This c-statistic was similar to that of the longer FRAX risk score (c = 0.76 for hip fractures), suggesting that the risk factors family history of fracture, daily alcohol use, and presence of RA can be omitted from the risk score. Our risk score predicted fragility fractures to a modest extent with an observed c-statistic of 0.71. This may be because of the fact that the outcome measure fragility fractures represent a combination of different fractures with different relationships to osteoporosis and aging.(36)

The cut-off point that we used to define women at high risk was based on recent guidelines that recommended the use of a 10-yr hip fracture risk of 5–10% or more as threshold for having a high risk.(9–12) We decided that the cut-off point must be somewhat higher in the oldest age group because otherwise all women ≥80 yr of age will be defined to have a high risk, and bisphosphonates may have less effect in persons ≥80 yr of age.(37) In this study, women 60–69 yr of age were defined as having a high risk with a cut-off point of 6%, whereas the cut-off point was 15% in women 70–79 yr of age and 22% in women ≥80 yr of age. The most appropriate cut-off point depends on the relative costs of case-finding and treatment to prevent fractures.

Our study also has some limitations. First, in LASA, the first 3 yr of the fracture follow-up were prospective, but in the second 3 yr, fractures were asked retrospectively. This may have led to an underestimation of fractures. Moreover, the registration of fractures was based on self-report. Although this method has been shown to be reliable,(38) there may have been persons who had a fracture that they never reported. This may also have led to an underestimation of the studied associations. Third, although this model has internally been validated, it has not yet been externally validated in another sample of community-dwelling elderly. Fourth, the respondents of both study samples are a selective group of relatively healthy older Dutch women because the frailest respondents were not able to participate at baseline. Therefore, these findings cannot automatically be generalized to men, frail elderly, and those living in nursing homes. Finally, data on prior fracture, use of a walking aid, and presence of RA were collected using different questions in the Rotterdam and LASA Studies. In the Rotterdam Study, prior fracture was assessed as prior fracture in the past 5 yr, whereas in LASA, it was assessed as prior fracture after age 50. In the Rotterdam Study, the use of a walking aid was assessed by asking whether a respondent has a walking aid or not. In the LASA Study, the researchers observed whether a respondent used a walking aid during a walking performance test.

In conclusion, the described risk score that is based on risk factors proposed in current guidelines is a practical, simple tool for the identification of older women with a high risk of osteoporotic fractures in primary care. Future studies are needed on validation and implementation of the score.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The authors thank Caspar Looman for help with data analysis and Jan Poppelaars, Marleen van der Horst, and Greetje Asma for help in data collection, management, and processing. This study was funded by the Efficiency Program of Zon-MW, The Hague, The Netherlands. This study is partly based on data collected in the context of the LASA, which is largely funded by the Ministry of Health, Welfare, and Sports of The Netherlands.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES
  • 1
    Cooper C, Campion G, Melton LJ III 1992 Hip fractures in the elderly: A world-wide projection. Osteoporos Int 2: 285289.
  • 2
    Melton LJ III , Chrischilles EA, Cooper C, Lane AW, Riggs BL 1992 Perspective. How many women have osteoporosis ? J Bone Miner Res 7: 10051010.
  • 3
    Lips P 1997 Epidemiology and predictors of fractures associated with osteoporosis. Am J Med 18: 3S11S.
  • 4
    Cooper C, Atkinson EJ, Jacobsen SJ, O'Fallon WM, Melton LJ III 1993 Population-based study of survival after osteoporotic fractures. Am J Epidemiol 137: 10011005.
  • 5
    Nevitt MC, Ettinger B, Black DM, Stone K, Jamal SA, Ensrud K, Segal M, Genant HK, Cummings SR 1998 The association of radiographically detected vertebral fractures with back pain and function: A prospective study. Ann Intern Med 128: 793800.
  • 6
    Philips S, Fox N, Jacobs J, Wright WE 1988 The direct medical costs of osteoporosis for American women aged 45 and older, 1986. Bone 9: 271279.
  • 7
    Anonymous 1998 Osteoporosis: Review of the evidence for prevention, diagnosis, and treatment and cost-effectiveness analysis. Osteoporos Int 8 (Suppl 4): S7S80.
  • 8
    Cummings SR, Black DM, Thompson DE, Applegate WB, Barrett-Connor E, Musliner TA, Palermo L, Prineas R, Rubin SM, Scott JC, Vogt T, Wallace R, Yates AJ, LaCroix AZ 1998 Effect of alendronate on risk of fracture in women with low bone density but without vertebral fractures: Results from the Fracture Intervention Trial. JAMA 280: 20772082.
  • 9
    Kwaliteitsinstituut voor de Gezondheidszorg CBO 2002 Osteoporose Tweede Herziene Richtlijn. Van Zuiden, Utrecht, The Netherlands.
  • 10
    NHG Standaard Osteoporose 2005 Eerste herziening. Huisarts Wetenschap. 48: 559570.
  • 11
    Kanis JA, on behalf of the World Health Organisation Scientific Group 2007 Assessment of Osteoporosis at the Primary Health Care Level. WHO Collaborating Centre for Metabolic Bone Diseases, University of Sheffield, Sheffield, UK.
  • 12
    World Health Organization 2007 Scientific Group on the Assessment of Osteoporosis at the Primary Health Care Level. Summary Meeting Report. World Health Organization, Geneva, Switzlerand.
  • 13
    Kanis JA, Johnell O, De Laet C, Johansson H, Oden A, Delmas P, Eisman J, Fujiwara S, Garnero P, Kroger H, McCloskey EV, Mellstrom LJ, Pols H, Reeve J, Silman A, Tenenhouse A 2004 A meta-analysis of previous fracture and subsequent fracture risk. Bone 35: 375382.
  • 14
    Kanis JA, Johansson H, Oden A, Johnell O, De Laet C, Eisman J, McCloskey EV, Mellstrom LJ, Melton LJ III , Pols HA, Reeve J, Silman AJ, Tenenhouse A 2004 A family history of fracture and fracture risk: A meta-analysis. Bone 35: 10291037.
  • 15
    De Laet C, Kanis JA, Oden A, Johansson H, Johnell O, Delmas P, Eisman JA, Kroger H, Fujiwara S, Garnero P, McCloskey EV, Mellstrom LJ, Melton LJ III , Meunier PJ, Pols HAP, Reeve J, Silman AJ, Tenenhouse A 2005 Body mass index as a predictor of fracture risk: A meta-analysis. Osteoporos Int 16: 13301338.
  • 16
    Kanis JA, Johnell O, Oden A, Johansson H, De Laet C, Eisman JA, Fujiwara S, Kroger H, McCloskey EV, Mellstrom D, Melton LJ, Pols H, Reeve J, Silman AJ, Tenenhouse A 2005 Smoking and fracture risk: A meta-analysis. Osteoporos Int 16: 155162.
  • 17
    Kanis JA, Johansson H, Johnell O, Oden A, De Laet C, Eisman JA, Pols H, Tenenhouse A 2005 Alcohol intake as a risk factor for fracture. Osteoporos Int 16: 737742.
  • 18
    Hofman A, Grobee DE, de Jong PT, van den Ouweland FA 1991 Determinants of disease and disability in the elderly: The Rotterdam Elderly study. Eur J Epidemiol 7: 403422.
  • 19
    Deeg DJH, Knipscheer CPM, van Tilburg W 1993 Autonomy and Well-Being in the Aging Population: Concepts and Design of the Longitudinal Aging Study Amsterdam. Netherlands Institute of Gerontology, Bunnik, The Netherlands.
  • 20
    Smit JH, de Vries MZ 1994 Procedures and results of the field work. In: DeegDJH, Westendorp-deSerière M (eds.) Autonomy and Well-Being in the Aging Population I: Report From the Longitudinal Aging Study Amsterdam 1992–1993. VU University Press, Amsterdam, The Netherlands, 713.
  • 21
    Pluijm SMF, Smit JH, Tromp EA, Stel VS, Deeg DJH, Bouter LM, Lips P 2006 A risk profile to identify community-dwelling elderly at high risk for recurrent falling: Results of a three-year prospective study. Osteoporos Int 17: 417425.
  • 22
    Kuchuk NO, van Schoor NM, Pluijm SMF, Smit JH, de Ronde W, Lips P 2007 The association of sex hormone levels with quantitative ultrasound, bone mineral density, bone turnover and osteoporotic fractures in older men and women. Clin Endocrinol (Oxf) 67: 295303.
  • 23
    Kriegsman DM, Penninx BW, van Eijk JT, Boeke AJ, Deeg DJ 1996 Self reports and general practitioner information on the presence of chronic diseases in community-dwelling elderly. A study on the accuracy of patients' self-reports and on determinants of inaccuracy. J Clin Epidemiol 49: 14071417.
  • 24
    Pluijm SMF, Steyerberg EW, Kuchuk NO, Rivadeneiro F, Looman CWN, van Schoor NM, Koes B, Mackenbach JP, Lips P, Pols HAP 2009 Practical operationalizations of risk factors for fracture in older women: Results from two longitudinal studies. J Bone Miner Res 24: 534542.
  • 25
    Harrell FE, Jr Lee KL, Matchar DB, Reichert TA 1985 Regression models for prognostic prediction: Advantages, problems, and suggested solutions. Cancer Treat Rep 69: 10711077.
  • 26
    Rubin DB, Schenker N 1991 Multiple imputation in health-care databases: An overview and some applications. Stat Med 10: 585598.
  • 27
    Harrell FE Jr , Lee KL, Mark DB 1996 Multivariable prognostic models: Issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 15: 361387.
  • 28
    Kanis JA, Borgstrom F, de Laet C, Johansson H, Johnell O, Jonsson B, Oden A, Zethraeus N, Pleger B, Khaltaev N 2005 Assessment of fracture risk. Osteoporos Int 16: 581589.
  • 29
    Kanis JA, Oden A, Johnell O, Johansson H, De Laet C, Brown J, Burckhardt P, Cooper C, Christiansen C, Cummings S, Eisman JA, Fujiwara S, Glüer C, Goltzman D, Hans D, Krieg MA, La Croix A, McCloskey E, Mellstrom D, Melton LJ III Pols H, Reeve J, Sanders K, Schott AM, Silman A, Torgerson D, van Staa T, Watts NB, Yoshimura N 2007 The use of clinical risk factors enhances the performance of BMD in the prediction of hip and osteoporotic fractures in men and women. Osteoporos Int 18: 10331046.
  • 30
    Siris E, Delmas PD 2008 Assessment of 10-year absolute fracture risk: A new paradigm with world-wide application. Osteoporos Int 19: 383384.
  • 31
    Hooyman JR, Melton LJ III , Nelson AM, O'Fallon WM, Riggs BL 1984 Fractures after rheumatoid arthritis. A population-based study. Arthritis Rheum 27: 13531361.
  • 32
    Kehoe R, Wu SY, Leske MC, Chylack LT Jr 1994 Comparing self-reported and physician-reported medical history. Am J Epidemiol 139: 813818.
  • 33
    Cummings SR, Nevitt MC, Browner WS, Stone K, Fox KM, Ensrud KE, Cauley J, Black D, Vogt TM 1995 Risk factors for hip fracture in white women. Study of Osteoporotic Fractures Research Group. N Engl J Med 332: 767773.
  • 34
    Huang Z, Himes JH, McGovern PG 1996 Nutrition and subsequent hip fracture risk among a national cohort of white women. Am J Epidemiol 144: 124134.
  • 35
    Burger H, van Daele PL, Algra D, Hofman A, Grobbee DE, Schütte HE, Birkenhäger JC, Pols HA 1994 Vertebral deformities as predictors of non-vertebral fractures. BMJ 309: 991992.
  • 36
    Ferreira-González I, Permanyer-Miralda G, Busse JW, Bryant DM, Montori VM, Alonso-Coello P, Walter SD, Guyatt GH 2007 Methodologic discussions for using and interpreting composite endpoints are limited, but still identify major concerns. J Clin Epidemiol 60: 651662.
  • 37
    McClung MR, Geussens P, Miller PD, Zippel H, Bensen WG, Roux C, Adami S, Fogelman I, Diamond T, Eastell R, Meunier PJ, Reginster JJ 2001 Effect of risedronate on the risk of hip fracture in elderly women. Hip Intervention Program Study Group. N Engl J Med 344: 333340.
  • 38
    Nevitt MC, Cummings SR, Browner WS, Seeley DG, Cauley JA, Vogt TM, Black DM 1992 The accuracy of self-report of fractures in elderly women: Evidence from a prospective study. Am J Epidemiol 135: 490499.