BMD and Fracture Risk

Many epidemiological studies over the last decade have helped to define the relationship between BMD and fracture risk. Some have been cross-sectional, others prospective. All have shown similar results. The largest prospective studies have reported the relative risk of fracture using logistic regression models. These relate the risk to the Z-score, in other words, to the standard deviations below the mean adjusted for age. The relative risk of a fracture is from 1.5 to 2.5 for each standard deviation below the age-matched mean. Often the Z-score and the T-score are confused. A 65 year old woman with a T-score of -1 does NOT have a risk twice that of an average 65-year-old woman! Her BMD would be a little above average; thus, she has no increase in relative risk. On the other hand, her absolute risk would be higher than that of an average 25-year-old woman. I think it is easier to relate the bone density to the absolute risk of fractures. Trying to figure out the absolute risks is difficult because many investigators don't report their data in sufficient detail, but the following graphs are my best attempts: 

seehttp://courses.washington.edu/bonephys/opbmd2.html for the graphs and more material which concludes by saying:

These graphs all show that fracture risk is definitely related to bone density. They also show that other factors  are even more important: age and presence of a vertebral fracture. There are still more risk factors that are  independent of the bone density, which are discussed in the section about fracture risk.

Bone densitometry is widely used in osteoporosis clinics to identify people at increased risk of fracture.  Terence Wilkin and Devasenan Devendra believe that evidence for the efficacy of bone densitometry is weak, but Jan Dequeker and Frank P Luyten argue that their interpretation of the evidence is too narrow and that screening high risk patients is cost effective

Osteoporosis: Assessing and Using Risk Factors for Fracture
Michael R. McClung, MD, Director of the Oregon Osteoporosis Center, Portland, Oregon
 [New Dimensions in Osteoporosis 1(1):2-8, 1999. © 1999 Medical Information Services, Inc.]
 http://www.medscape.com/MIS/osteoporosis/1999/v01.n01/ndio0101.mccl/ndio0101.mccl-01.html
 Educational Objectives
 Upon completion of this educational activity, physicians should be able to:

1. Recognize the role of fracture risk in osteoporosis diagnosis and treatment. 
2. Identify clinical risk factors for fracture. 
3. Understand the relation between bone mineral density (BMD) and fracture risk. 
4. Identify appropriate use of BMD testing in clinical practice 

Osteoporosis is now recognized as a disorder of fracture risk characterized by low bone mass and abnormal skeletal microarchitecture.This article will review the relationship between various risk factors and fracture and will address the clinical utility of risk factors in postmenopausal women.

• The relative risk for any fracture is about 1.5 for a 1 SD decrease in BMD at any site. 
• The prediction of site-specific fracture risk, especially for hip fractures, is improved by measurement at that site.
• It is not clear that measurement at multiple sites actually improves fracture risk assessment.
• Data suggest that BMD is a predictor of fracture risk across a broad spectrum of age, but becomes relatively less important with advancing age.EPIDOS is a large prospective study of French men and women over age 75. In that study the relative risk of hip fracture for a 1 SD change in hip density was

4.4 in women below age 80

2.5 in women older than 80 years 

only 1.6 in women 85 years and older.

• BMD measurements have also been shown to be predictive of fracture risk over at least a 25-year interval. The longer the interval, the less important today's bone density value is, relative to the effect of bone loss, for future BMD and fracture risk.

Interpretation of Bone Density Tests

BMD values are "normally" distributed in a group of young adults i.e.. 

• 67% of young adults will fall within 1 SD of the mean (T score between -1 and +1)
• 95% will have values within 2 SD of the mean (T score between -2 and +2). 
• about 16% of young normal adults would be categorized as having low bone density according to the WHO criteria,  even though they are within the true normal reference range. This emphasizes the point that a low BMD value does not necessarily mean that bone loss has occurred. For some individuals, a T score value between -1 and -2 is their peak bone mass.

• 16.4 in the women with osteoporosis by the WHO criteria (-2.5SD)
•  5.4 if they had low bone mass (-1 to -2SD)
•  1.1 if their T scores were </= -1. 

Other Risk Factors for Fracture

• #1 is age. (Fracture risk doubles for each 5 - 10 years of age.)

• the presence of previous fractures,
• indices of bone metabolism, other skeletal factors
• falls and other injuries - risks for which include
• sedative drug use
• hypotension 
• poor balance 
• weakness
• visual or neurologic impairment
• medical problems that predispose to bone loss or injury - including. 
• chronic steroid use
• suppressive doses of thyroid hormone
• intestinal malabsorption
• clinical risk factors predictive of fracture (even after adjusting for age and BMD) include
• history of smoking or hyperthyroidism
• low body weight(<58kg or a history of weight loss)
• pulse rate >80bpm

For the clinician, the value of considering these risk factors is 2-fold. 

• Some of these factors are amenable to therapy that could result in decreased fracture risk. 
• these risk factors, even those that cannot be changed, identify individuals for whom pharmacologic therapy is appropriate on the basis of high fracture risk.

In prospective studies, combining risk factors provides better stratification of fracture risk among individuals; clinical risk factors and BMD can be combined to enhance risk prediction. 
Those individuals with

• multiple clinical risk factors for fracture were at increased risk for hip fracture compared to those with few clinical risk factors, irrespective of their bone density.
• with low BMD were at higher fracture risk than those with higher BMD, regardless of the number of clinical risk factors. 
• patients with both low BMD and multiple clinical risk factors were at the greatest risk for fracture.

Critics have pointed out that markers of bone turnover are "poorly predictive of bone mineral density ... and cannot be used to diagnose osteoporosis or to select patients for subsequent densitometry," but this is to miss the point. If the modest gain in bone density seen with treatment is insufficient to account for the substantial reduction in fracture risk, a state of high bone turnover, rather than its prevailing mass, may be the responsive element in fracture prevention, no matter at what age it is encountered. The clinical implications are important, as there is no evidence that bone density identifies those people who will sustain a fracture, but abundant evidence that restoring bone turnover to normal values universally reduces the risk. A switch in emphasis from bone density, which declines irreversibly, to bone turnover, which rises, but is fully reversible, makes reducing the risk of fracture a viable consideration at any age after the menopause. 

Women with hip fractures had lower levels of 25-hydroxyvitamin D than women without osteoporosis admitted for elective joint replacement (P=.02) and than women with osteoporosis admitted for elective joint replacement (P=.01) (medians, 32.4, 49.9, and 55.0 nmol/L, respectively; comparisons adjusted for age and estrogen intake)