WHO calculator revised to use g/cm2

Click to link to the World Health Organization FRAX calculator. Look for the pulldown menu called "calculation tool". You will need to know the manufacturer of the bone density machine (Hologic, Lunar or Norland) and the actual bone density in grams per square centimeter for the femoral neck.

This finally will eliminate all the problems with T-scores!

The FRAX calculator uses femoral neck, although recent studies from Canada show that the total hip is better able to predict the risk of a fracture. The reason for using the femoral neck is that many of the older studies (which have the necessary longitudinal followup) did not include the total hip. So if the total hip bone density is more than one standard deviation better than the femoral neck, and the region of interest on the image is in the correct place, then the fracture risk will be better than predicted with FRAX.

The WHO calculator became available in February 2008. Here are some things to note:

• Swedish risks are higher than US for the same T-score.
• People with no risk factors have lower fracture risks than shown in my old calculator.
• The risk between family history and fracture depends on the age.
• Combinations of over 3 factors give worse predictions in my old calculator.
• The WHO calculator does not give spine fracture risks.

I'll leave this older calculator here for a little while, especially to help with spine fracture prediction which was based on the SOF study. I have not updated it to match the FRAX calculator.

The latest version corrects a small error when there were multiple risk factors. The calculator has included the risk factors from several meta-analyses by the WHO (Kanis, 2005) which apply to both men and women. A recent paper from these investigators (Kanis, 2007) has validated these clinical fractures with several very large cohorts. When there are several risk factors, they may interact with each other, so simply multiplying the base rate time the relative risks will only be an estimation. Furthermore, the base rates are taken from the paper by Johnell, which are from Sweden.

Alcohol: A "standard drink" in the USA is 14 grams of ethanol. The following drinks contain this amount: one can (12oz) beer, one small glass (5oz) wine, 1 jigger (1.5oz) spirits. The alcohol content varies among brands so these are approximate. In the UK, alcohol is expressed by "units" and each unit is 8 grams, so that a "drink" is about 2 units.
Weight: The Black index used weight less than 125 pounds, which applies to women. The WHO used Body Mass Index, which also depends on height, and this risk should be clicked if BMI is less than 21. Details about clinical risk factors are in the page about clinical risk factors.

The calculator now includes the absolute fracture risk as recommended by the Canadian Association of Radiologists guidelines (Siminoski K). These have been modified slightly because I did not include bone density of the radius nor of the spine in men.

For women whose spine bone density is lower than the hip, the vertebral fracture risk is more closely predicted by the spine measurement (Fink, ASBMR 2005 abstract), so it is now possible to enter both hip and spine measurement for women. There are more details in the new page about discordance.

With all these new changes, I hope I have not made any errors. I can't guarantee that this is bug-free, so let me know if you find a problem. I would like to acknowledge Dr. Dean Kramer who has essentially been a beta tester and discovered several errors in the calculations. Send me an email message.

The T and Z scores and standardized bone density values are based on the NHANES data published by Looker. The SOF (Study of Osteoporotic Fractures) data was from several papers including Cummings, 2002 and Black. The intervention thresholds for cost-effectiveness were taken from Kanis. The meta-analysis by Johnell was used to predict the 10-year hip fracture risks. There are many details in the page about fracture risks.

These sources all used slightly different age ranges or bone density ranges, so the values have been extrapolated. For persons of African ancestry, the fracture risk calculations assumed that the relative risks for the bone density and fractures were the same as for Caucasian persons, more data about this are needed.

The risk factors may interact with each other and they may be more important at different ages. As more work is done, this calculator will need to be updated to take these interactions into account. Also, the cost-effectiveness will be changing as new drugs are available, as costs change, and as complications are defined.

References