Your DXA Questions Answered by Larry Jankowski CBDT - Part I
Biographical Sketch – Larry Jankowski, CBDT
Larry Jankowski is the chief DXA technologist and a research study coordinator for Illinois Bone and Joint Institute, a physician group practice with 19 locations, 81 orthopedic surgeons, and 7 rheumatologists in Chicago, and its north and westerns suburbs. He has been performing DXA scans for over 25 years. He is certified by the International Society for Clinical Densitometry (ISCD) since 1996 as a CBDT. He also is an instructor for the ISCD/IOF Osteoporosis Essentials courses in both the United States and internationally. Larry was the first DXA technologist elected to the ISCD Board of Directors in 1996, and has held the office of treasurer from 1997 to 2000, and again from 2010-2013. He served as an expert panelist for several ISCD Position Development Conferences, and has presented posters at bone meetings including the ISCD, NOF, ASBMR, and AAOS.
He is also on the editorial board for the Journal of Clinical Densitometry, is a member of the ISCD Facility Accreditation Council, and a co-editor of the ISCD “Case of the Month” e-learning project and also volunteers as a moderator for the NOF/Inspire Osteoporosis Support Group.
In his spare time, which he claims he has less and less of, he enjoys fishing, carpentry/woodworking, baseball, and being a grandfather. He is still happily married to his wife Carol for 32 years (and counting).
Welcome Mr. Jankowski!
Q: When should you have a DXA scan?
A: No test, not even a DXA scan, should be performed unless the results of the test would change the way the patient is being managed. For example, a healthy 30-year-old woman who exercises regularly, eats a diet that is rich in calcium and has a normal vitamin D level, would not benefit from a bone density test. She is already doing all the things that a 30-year-old person should be doing. None of the medications for osteoporosis, even if it turned out she had low bone density, are approved for use in young healthy individuals, and so exercise, adequate calcium and vitamin D would be the recommendations from the results of a test that turned out low. It would also be in the results of a test that turned out to be perfectly normal. In other words, the results of the test would not change the recommendations for the patient, and the test would be considered unnecessary.
Having a baseline scan done at the time of menopause, may make more sense. But only if the patient has some other risk factors suggest that they are more likely to have low bone density than not. These risk factors would include a current smoker, a strong family history of hip fracture in the mother or father, long-term use of glucocorticosteroids, for women that have a history of amenorrhea lasting more than six months, or anorexia or bulimia, or taking hormonal therapy (e.g. Lupron) or chemotherapy drugs for breast cancers called aromatase inhibitors.
Age is also a risk factor for developing osteoporosis. If you have not had a bone density test and you are a woman age 65, it is included in the welcome to Medicare exam for that reason. Men aged 75 and older who have not had a bone density test would be a sufficient risk for osteoporosis to warrant a bone density test as well.
Many professional medical societies and patient advocacy groups have position statements indicating who should be tested and when, including the National Osteoporosis Foundation, the International Society for Clinical Densitometry, the American Academy of Clinical Endocrinologists, the American College of Rheumatology, the North American Menopause Society, the Centers for Medicare and Medicaid Services, etc. While each can recommend who should get a scan, often it is who is going to pay for a bone density test (e.g. CMS) that determines who gets tested.
Q: What is the margin of error on a DXA scan? Is this important to know?
A: Every instrument that measures a physical quantity has some sort of measurement error, what you are referring to as margin of error. DXA scanning is no different. To experience the phenomenon firsthand, put your arm in an automated blood pressure cuff at the local drugstore and without moving your arm, press the button several times in a row and you will find that you get several different blood pressure readings. If you were to calculate the average systolic and the average diastolic value, as well as the highest and lowest value outside of the average, you would have some idea of the variability or uncertainty of a physical measurement.
DXA scanning facilities that adhere to the ISCD official positions on precision, calculate their reproducibility or precision error, as well as something called the “least significant change” or LSC. They do so by scanning a number of volunteers twice in a row getting them on and off table between each set of measurements, to simulate what happens when a patient returns years later. The average difference between the first and second measurements, averaged over 30 to 50 patients, is used to mathematically estimate the precision error and LSC values for that scanner, and that technologist. If you do not know what the LSC is at a particular facility, you cannot make any quantitative statement on whether a patient's bone density has changed between two measurements on that device. So it is a very important thing to know.
Just as an aside, most of the error in DXA measurements is caused by the differences in positioning of the bone being scanned in the x-ray beam, not the machine itself, unless the machine is malfunctioning. Technologist who do not take the time to properly position the patient, and who do not stop and restart a scan that is not exactly the same positioning as a baseline scan, can be the largest source of precision error in the measurements. It is extremely difficult to achieve identical positioning of the bones on scans being done the same day, and it is almost impossible to achieve exactly the same positioning in scans done several years apart. Unless the DXA facility has set up a rigorous quality assurance and technologist training program, and checks their precision and LSC values against a set of minimum standards such as those established by the International Society for Clinical Densitometry, accurate monitoring of changes in bone density is not possible.
The better the precision, the smaller the LSC, and the smaller changes one can measure with statistical confidence over time. In my 27 years of experience doing bone density scans and reviewing the scans from other DXA facilities, those that have only a few dedicated technologists who perform DXA scanning as their primary or only job, have much better precision than centers who have many technologist each rotating for a short period of time through the DXA lab, such as a large teaching hospital, or scan operators who are busy doing many other tests such as a office nurse or nursing assistant in a small doctor's practice.
Q: What is the difference between a Z-score and a T-score? What population of patients does each apply to and why?
A: In order to understand what a T-score or a Z-score is you have to understand what is known in statistical terminology as a normal distribution. The lay term for this is a bell curve. Just about everything we measure in nature, including bone density, forms a bell shaped distribution. There are more people closer to the average - the peak of the bell curve, and fewer and fewer people as you get out to values either higher or lower than the average, in what are sometimes referred to as the “tails” of the curve. Retailers are very familiar with bell shaped curves and use it to determine how much of a particular size shoe they have to keep in stock for example. If the average shoe size for men is a 9 1/2, then they will have more shoes of that size and a little bit larger and smaller than that size and only a few in very small sizes or very large sizes in order to match stock on hand to meet the demand.
A standard deviation or SD is a statistical term that describes the width of a bell curve. It is called standard because plus or minus one standard deviation will always encompass the middle 68% of the bell curve population. And plus or minus 2 standard deviations will always encompass about 95% of a normal distribution. T-scores and Z. scores are in SD units. The wider the bell curve, the larger the standard deviation for that normal distribution.
Both T-scores and Z-scores are in units of SD’s. The only difference is that a T-score compares you against a young normal distribution of bone density values (regardless of your age), and a Z-score compares you against an age, sex, and race matched normal distribution of values.
In 1994, a working group of the World Health Organization defined osteoporosis as having a T-score that was -2.5 standard deviations below the mean of a young normal population of bone density. A T-score between -1 and -2.5 was called "osteopenia". Anything above a T-score of minus one was considered normal. It was never meant to be used to make a diagnosis in an individual. It was only as a tool for countries to use to estimate how many patients would suffer and osteoporotic fracture, so that they could calculate how much healthcare resources to allocate to the problem. The medical community, lacking anything better, began using the T-score definition to diagnose osteoporosis in individual patients.
The reason the World Health Organization chose the cut points of -2.5 was that the number of women over the age of 65 who would suffer an osteoporotic fracture of the hip was approximately equal to the number of people in that same population who would have been found to have a T-score at or below -2.5 at the hip. They wouldn't necessarily be the same people, as there is a lot of overlap between those who fracture and those who have T-score below -2.5. Thus by simply scanning a small representative a random sample of older people in that country, calculate what percentage had T-scores below -2.5 at the hip, and then could predict the percentage of that population that would go on to have hip fractures.
But fracture risk in adult’s increases as BMD falls. If we adjust for age, sex and tried to use Z-scores then as people aged, the Z-score would not change even though fracture risk increased. That is why T-score are used for diagnosis in adults past the age of peak bone mass, which occurs between ages 18 and 30, depending on which bone you are looking at. However, in patients who are young and have not yet reached peak bone mass, the Z-score is the more appropriate value to use. The average Z-score of a newborn is zero, by definition. But this BMD value calculates to a T-score around -8.0. Clearly an average newborn does not have severe osteoporosis, which would be suggested by such a low T-score.
The ISCD recommends using T-score is in postmenopausal women and men over the age of 50 and Z- scores for anyone younger than that. Between ages 30 and 50, there is little difference in the T-score and Z-score as we tend to reach peak bone mass and stay there until women reach menopause and men get to age 65-70.
Q: Can a DXA scan give an accurate score where the patient has one or more of the following: joint replacement, arthritis, fracture or spinal hardware?
A: While it is possible to get an accurate measure of the bone density in what little bone is left surrounding a joint replacement, or in the spine and vertebral bodies where screws and metal plates and rods are present, or where severe arthritic changes have altered the normal architecture of the bone, the readings, while accurate, are not clinically useful as they have lost the relationship of BMD to fracture risk, and trying to compare them to normal values, where patients with these conditions or implants were not included in the construction of the normal data would be like trying to compare apples to oranges.
As a DXA technologist, I am trained to identify artifacts and degenerative diseases, and whenever possible to exclude those from the results that the interpreting physician will be using to generate the DXA report. It may mean only scanning one hip if the other had been replaced, the spine is full of arthritis, and the patient had recently broken both wrists. Or it may mean deleting one vertebral body that has developed a spur (called an osteophyte) since the last scan.
Q: We receive several scores on a DXA print-out; what does each score mean?
A: If you are a postmenopausal woman or a man over the age of 50 you should have both a T-score and a Z-score included for each of the regions of the skeleton that were measured. Most commonly it will include a T-score and Z-score for the lumbar spine region of L1-L4, (including individual levels and the total of all four), the femoral neck and total hip, and in some cases for the one-third distal (furthest third away from the elbow), radius bone in the forearm and perhaps the greater trochanter of the hip. The lowest T-score of L1-L4, the femur neck, or the total hip, (or the 1/3 distal radius if either the spine or hip cannot be measured) is what should be used to give you your diagnosis. You should not have more than one diagnosis. Oftentimes I see reports which state the patient has “osteopenia in the spine, osteoporosis at the femoral neck, and normal at the total hip. “ Or they cherry-pick one vertebral body out of the average of the four. This seems silly, as we do not have drugs that can treat only the femoral neck, or a single vertebral level, but ignore the total hip or the forearm because they were in the normal range.
More recently a new score called FRAX (www.shef.ac.uk/FRAX)has begun to appear on DXA scan results. Like the T-score definition of osteoporosis, FRAX was developed by a working group of the World Health Organization in an attempt to improve upon its older definition and be more clinically useful in treating individual patients. It takes into account not only the bone mineral density at the neck of the femur for the prediction of hip fracture risk when such a value is available, but also a number of other risk factors independent of bone density that also can increase or decrease one's risk of fracture. Things like current smoking, your current height and weight, a family history of hip fracture in a mother or father, whether or not you have rheumatoid arthritis, whether or not you have been exposed to high doses of prednisone or other glucocorticosteroids, as well as your age and whether or not you've already experienced a fragility fracture are all strong and independent risks for fracture.
The FRAX calculations provide two percentages. One is your 10 year risk of having a major osteoporotic fracture, and the other is your 10 year risk of having a hip fracture in particular. Each country has its own FRAX calculator which takes into account things like life expectancy from actuarial tables, and hip fracture incidence rates by age from medical registries or observational population studies, in addition to the individual risk factors mentioned above. It allows countries to determine what level of risk is cost-effective for them to treat a patient for their osteoporosis. If you live in a very poor nation, the risk of fracture must be higher in order to justify diverting health resources to treat the disease. If you live in a much more affluent nation, then the cost of treatment versus the cost of the fracture is typically a lower threshold, since resources do not have to be diverted to the more serious diseases common in developing nations such as cholera, malaria, AIDS, ebola, etc.
In the United States for example, the National Osteoporosis Foundation has determined through economic modeling, that a 10 year risk of major osteoporotic fracture greater than 20%, and/or a 10 year risk of hip fracture greater than 3%, is high enough to justify the cost of osteoporosis treatment with pharmaceuticals to prevent fractures.
Thank you Mr. Jankowski for a very informative interview. We look forward to Part II.