Info

a Per SD decrease in BMC b Per SD decrease in BMD c Per SD decrease in RA units From refs. 2,4,7-10.

a Per SD decrease in BMC b Per SD decrease in BMD c Per SD decrease in RA units From refs. 2,4,7-10.

calculate the relative risk for the patient in question is clearly an extrapolation of the data that may not be appropriate. In addition, relative risk values are generally calculated as age-adjusted values. Age adjustment means that the effect of age has been accounted for or eliminated statistically such that only the effect of bone density on risk remains. These considerations profoundly affect the densitometrist's ability to use relative risk data in clinical practice.

The calculation of a patient's relative risk for various types of fracture using standard scores on the densitometry printout and published relative risk data is not difficult. The relationships between absolute risk, relative risk, and the magnitude of the SD decline in bone density that were discussed in Chapter 3 are the key to this calculation and its interpretation. For example, in the spine bone density study shown in Fig. 10-1, the L1-L4 BMD is 0.814 g/cm2. The T-score of -2.12 indicates that the L1-L4 BMD is 2.12 SD below the average peak BMD of the young adult. The z-score of -0.08 indicates that the L1-L4 BMD is 0.08 SD below the average BMD predicted for this woman's age. How does a physician calculate the global fracture relative risk or site-specific spine fracture relative risk? The age-adjusted increases in relative risk for fracture from Melton et al. (4) can be used to calculate these values. Melton et al. found that for each SD decline in bone density when measured at the spine, the increase in relative risk for any type of low to moderate trauma fracture was 1.5. Therefore, the relative risk for global fracture in this patient compared to the individual who still has an average peak bone density, based on this measurement of bone density at the lumbar spine, is 1.5212 or 2.4. This is the increase in relative risk per SD decline in bone density raised to the power of the TT-score. Compared to the individual who has a spine bone density that would be predicted for the patient's age, her global fracture relative risk would be 1.5008 or 1.03. This is the increase in relative risk raised to the power of the z-score. Strictly speaking, neither of these calculations is correct, although it is the best that can be done with the data at hand.

Fig. 10-1. Hologic QDR PA lumbar spine study in a 70-year-old Caucasian woman. The L1-L4 BMD is 0.814 g/cm2. The T-score is -2.12 and the z-score is -0.08. The correct diagnosis is osteopenia. The patient also meets the NOF guidelines for prescription intervention in osteoporosis. Use of relative risk fracture data with either the T- or z-score will lead to very different impressions of this patient's fracture risk.

Fig. 10-1. Hologic QDR PA lumbar spine study in a 70-year-old Caucasian woman. The L1-L4 BMD is 0.814 g/cm2. The T-score is -2.12 and the z-score is -0.08. The correct diagnosis is osteopenia. The patient also meets the NOF guidelines for prescription intervention in osteoporosis. Use of relative risk fracture data with either the T- or z-score will lead to very different impressions of this patient's fracture risk.

The reason that neither calculation is correct is because the strict interpretation of the 1.5 relative risk for global fracture per SD decline in spine BMD from the Melton study (4) is that an individual in the Melton study has a 1.5-fold higher risk for global fracture than another individual in the same study who has a BMD that is 1 SD higher. There is nothing here that a densitometrist can really use to quantify fracture risk in his or her individual patient. Because the data are age-adjusted, and the BMD study mean and SD are unique for the study population, the use of the individual patient's T- or z-score is not correct. The use of the patient's z-score is probably the better of the two standard scores on the bone density printout with which to use relative risk data but in this case, might clearly lead to confusion. The patient would be diagnosed as osteopenic using WHO Criteria2 and meet the NOF Guidelines for prescription intervention and yet the use of the z-score with relative risk fracture data might lead the physician to conclude that she is not at increased risk for fracture.

Relative risks for site-specific spine fracture prediction are calculated in a similar fashion. Using an increase in relative risk per SD decline in BMD of 2.2 for spine fracture when measured at the spine, the calculation would be 2.2212 or 2.20 08, depending on the comparison the physician wishes to make.

2 See Chapter 9 for a discussion of the WHO Criteria for the diagnosis of osteoporosis.

Global fracture relative risk data for BMD measurements made at other sites can be calculated using data from Melton et al. (4) as well as other authors (2,3,5). Similarly, the relative risk for a site-specific spine fracture risk prediction can be calculated from bone density measured at other sites using data from Melton et al. (4) and others (2,7,8). The data from Cummings et al. (9) are most commonly used for the calculation of relative risk for site-specific hip fracture risk prediction.