The Value of Bone Density Testing

Synopsis: An article just published in the online version of the British Medical Journal (BMJ 2009;338:b2266) concludes that bone mineral density (BMD) testing by DXA to monitor response to bisphosphonate therapy in older women is unnecessary and potentially misleading. The article was quickly picked up by WebMD Health News and Forbes within 24 hours and in all likelihood will circulate extensively through the lay press.

The ISCD finds major clinical value in BMD testing to monitor therapy for patients with osteoporosis and is concerned that misperceptions will be created by this article potentially adversely impacting quality of patient care. We are currently in discussions about how to provide accurate information to the lay public as well as the media and those in Congress.

If you are asked to comment on this paper, you might consider the following talking points:

  • The study looks at a controlled clinical trial that does not accurately reflect real world experience
    • Many common medical conditions were excluded from trial entry; most real world osteoporosis patients do not qualify for osteoporosis studies.
    • Clinical trial participants are generally a motivated group of individuals that may not reflect real world clinical practice. Consistent with this, compliance was extraordinarily good (96%) unlike usual patients where compliance is typically no more than 40%.
    • Precision assessment was not done in this trial
    • Patients who experienced rapid bone loss were excluded
  • The findings of a posthoc analysis of a single trial with one drug cannot be generalized to other drugs or other patient populations.
  • BMD testing using DXA is of value in monitoring response to therapy. Intervals between testing should be determined according to each patient’s clinical status, typically one to two years after initiation or change of therapy.
  • BMD testing remains the standard of care for the diagnosis of osteoporosis in those postmenopausal women who have not yet fractured.
  • BMD testing by DXA is the only BMD measurement that can be used with FRAX, and provides better assessment of fracture risk with FRAX than clinical risk factors alone.

On June 24th, the British Medical Journal released in its Online First version an article entitled “Value of routine monitoring of bone mineral density after starting bisphosphonate treatment: secondary analysis of trial data” by Bell, Hayden, Macaskill, Irwig, Craig, Ensrud and Bauer. (
The paper looks at data from the Fracture Intervention Trial (FIT 1 and 2), a large clinical trial of post-menopausal women randomized to alendronate vs placebo. The authors conclude that bone density testing by DXA to monitor treatment response was “unnecessary, and because of potential to mislead, best avoided”. An accompanying editorial by Juliet Compston ( also championed this approach. Also on June 24th, WebMD Health News ran a story about the study under the title “Do Bone Density Tests Help Monitor Osteoporosis? Study Shows Bone Density Tests May be Misleading for Patients Getting Bisphosphonates.” Towards the end of the WebMD article, Dr Robert Recker, current president of the National Osteoporosis Foundation was briefly quoted in rebuttal.

The BMJ paper has many significant flaws. It assumes that the results of a tightly controlled clinical trial can be generalized to apply to real world patients.

  • Compliance in this trial approached 96%, yet in actuality, less than 40% of patients are compliant with bisphosphonates.
  • The trial excluded patients with dyspepsia, prior peptic ulcer disease, and major medical problems; these are the very patients who in the real world are often prescribed drugs to treat osteoporosis.
  • When the FIT trial was undertaken in the early 1990s, precision testing to determine least significant change was not performed. Thus, one does not know what the least significant change was when looking at “within-person” (measurement related) variation reported in this study.
  • The dose of alendronate used for the first two years (5 mg a day) is not the currently recommended dose for treatment of osteoporosis.
  • Significantly, the trial itself had a mandatory drop out clause whereby patients who had significant bone loss measured on DXA (8% over 1 year, 10% over 2 years, 12% over 3 years) were terminated from the trial.

Thus, the lack of variability in response from patient to patient in this clinical trial was in part related to the specific exclusion criteria employed, the type and dose of medication used, the exceptionally good compliance and the enforced drop out for outliers. Based on this lack of variability, the authors incorrectly conclude that monitoring does not make sense in all older women on all bisphosphonates.

Obtaining a bone density study at some time after onset of treatment and comparing it to a baseline study performed at the same institution is of value. If a significant decline in bone density is identified which exceeds the least significant change, one can reasonably conclude that the patient is not compliant with the drug, the drug is not working and/or a secondary cause of metabolic bone disease has developed. This is indeed important information for both the doctor and the patient. In practice, patients will be less inclined to continue on a medication for 3 years or longer without having some measurable results. Multiple studies have demonstrated that increases in bone density while on bisphosphonate therapy do indeed correlate with reduction in fracture risk.

Recently, generic alendronate has become commercially available. As an inexpensive drug ($4/month or $10/ 3 month at Walmart) it has great appeal to those wishing to minimize healthcare costs. Yet, the FDA did not require generic alendronate to show comparable BMD results or fracture efficacy to branded alendronate (Fosamax). Recent studies have shown that solubility of generic alendronate differs greatly among the various manufacturers and improvement in bone density was less than for those on Fosamax. Armed with the erroneous conclusions from the BMJ paper, many would not bother to have their bone density checked to see if the generic alendronate they were taking was effective.

Most importantly, the message that some people may be left with is that bone density testing is not appropriate at any time including for diagnosis of the disease. The WebMD Health News article quotes Dr Compston who states that “bone mineral density testing is not a particularly good measure of fracture risk“. In fact, numerous studies have demonstrated that BMD testing by DXA is a better predictor of fracture risk than hypertension for stroke or hyperlipidemia for MI. In 2002, the United States Preventive Services Task Force recommended that all women over the age of 65 should be screened for osteoporosis by DXA testing. Numerous medical societies representing primary and specialty care as well as preventive medicine have published guidelines that recommend BMD testing by DXA for diagnosis and monitoring response to treatment. DXA testing is also included as a basic benefit in the Welcome to Medicare Exam. For post menopausal women who have not yet fractured, DXA testing remains the standard of care for diagnosis of the disease. Recent attempts to even further improve fracture risk prediction have incorporated DXA with other models of fracture risk (e.g. FRAX).

At a time when only 22% of Medicare women who have had an osteoporotic fracture have a DXA study or go on treatment, we need to do a better job recognizing and treating this serious disease. In 2010, it is estimated that we will spend over $20 billion dollars on osteoporosis related to complications of fractures. Twice as many people in this country have osteoporosis or low bone mass as have diabetes and more patients die from complications of hip fracture every year than die from breast cancer. The worst possible message that we could be sending at this time is that bone density testing is not of value for diagnosis or monitoring response to therapy.

Andrew Laster MD, FACR, CCD
President – International Society for Clinical Densitometry (ISCD)

Last modified: December 17, 2012