Development of prognostic nomograms for individualizing 5-year and 10-year fracture risks
- First Online:
- 457 Downloads
We have developed clinical nomograms for predicting 5-year and 10-year fracture risks for any elderly man or woman. The nomograms used age and information concerning fracture history, fall history, and BMD T-score or body weight.
Although many fracture risk factors have been identified, the translation of these risk factors into a prognostic model that can be used in primary care setting has not been well realized. The present study sought to develop a nomogram that incorporates non-invasive risk factors to predict 5-year and 10-year absolute fracture risks for an individual man and woman.
The Dubbo Osteoporosis Epidemiology Study was designed as a community-based prospective study, with 1358 women and 858 men aged 60+ years as at 1989. Baseline measurements included femoral neck bone mineral density (FNBMD), prior fracture, a history of falls and body weight. Between 1989 and 2004, 426 women and 149 men had sustained a low-trauma fracture (not including morphometric vertebral fractures). Two prognostic models based on the Cox’s proportional hazards analysis were considered: model I included age, BMD, prior fracture and falls; and model II included age, weight, prior fracture and fall.
Analysis of the area under the receiver operating characteristic curve (AUC) suggested that model I (AUC = 0.75 for both sexes) performed better than model II (AUC = 0.72 for women and 0.74 for men). Using the models’ estimates, we constructred various nomograms for individualizing the risk of fracture for men and women. If the 5-year risk of 10% or greater is considered “high risk”, then virtually all 80-year-old men with BMD T-scores <-1.0 or 80-year-old women with T-scores <-2.0 were predicted to be in the high risk group. A 60-year-old woman’s risk was considered high risk only if her BMD T-scores ≤-2.5 and with a prior fracture; however, no 60-year-old men would be in the high risk regardless of their BMD and risk profile.
These data suggest that the assessment of fracture risk for an individual cannot be based on BMD alone, since there are clearly various combinations of factors that could substantially elevate an individual’s risk of fracture. The nomograms presented here can be useful for individualizing the short- and intermediate-term risk of fracture and identifying high-risk individuals for intervention to reduce the burden of fracture in the general population.
KeywordsBone mineral density Epidemiology Fall Fracture risk Nomogram Osteoporosis Prior fracture
- 7.Wyatt JC, Altman DG (1995) Commentary: prognostic models: clinically useful or quickly forgotten? BMJ 311:1539–1541Google Scholar
- 10.Johnell O, Kanis JA, Oden A, Johansson H, De Laet C, Delmas P, Eisman JA, Fujiwara S, Kroger H, Mellstrom D, Meunier PJ, Melton LJ 3rd, O, Neill T, Pols H, Reeve J, Silman A, Tenenhouse A (2005) Predictive value of BMD for hip and other fractures. J Bone Miner Res 20:1185–1194PubMedCrossRefGoogle Scholar
- 11.Kanis JA, Johnell O, De Laet C, Johansson H, Oden A, Delmas P, Eisman J, Fujiwara S, Garnero P, Kroger H, McCloskey EV, Mellstrom D, Melton LJ, Pols H, Reeve J, Silman A, Tenenhouse A (2004) A meta-analysis of previous fracture and subsequent fracture risk. Bone 35:375–382PubMedCrossRefGoogle Scholar
- 16.National Osteoporosis Society (1998) Guidance on the prevention and management of corticosteroid induced osteoporosis. National Osteoporosis Society, Camerton, Bath, UKGoogle Scholar
- 23.Volinsky CT, Madigan D, Raftery AE, Kronmal RA (1997) Bayesian model averaging in proportional hazard models: assessing the risk of a stroke. Appl Statist 48:433–448Google Scholar
- 24.Raftery AE (1995) Bayesian model selection in social research. In: Marsden PV (ed) Sociological methodology. Blackwell, Cambridge, MAGoogle Scholar
- 29.Cox DR (1972) Regression models and life tables. J R Stat Soc (B) 34:187–220Google Scholar
- 30.Cox DR, Snell EJ (1989) Analysis of binary data. Chapman and Hall, London, U.KGoogle Scholar
- 32.R Development Core Team (2007) R: a language and environment for statistical computing. In. R Foundation for Statistical Computing, Vienna, AustriaGoogle Scholar
- 33.Harrell FE Jr, Margolis PA, Gove S, Mason KE, Mulholland EK, Lehmann D, Muhe L, Gatchalian S, Eichenwald HF (1998) Development of a clinical prediction model for an ordinal outcome: the World Health Organization Multicentre Study of Clinical Signs and Etiological agents of Pneumonia, Sepsis and Meningitis in Young Infants. WHO/ARI young infant multicentre study group. Stat Med 17:909–944PubMedCrossRefGoogle Scholar
- 46.Harrel FE (2001) Regression modeling strategies with applications to linear models, logistic regression, and survival analysis. Springer, New York, NYGoogle Scholar
- 52.Hosmer DW, Lemeshow S (1999) Assessing the fit of the model. In Apllied logistic regression. Wiley J, New York, pp 135–175Google Scholar