What Kind of Errors Can Happen With REMS and How Can They be Prevented?

Let us start off by first stating the obvious - that we are unabashed proponents of REMS technology. We are building successful bone health programs based on REMS. What started off as curiosity with a new and innovative technology led to an “a-ha!” moment - understanding the underlying principles of how sound waves interact with bone to provide an accurate measurement of BMD and an assessment of Bone Quality. Those out there who really understand the unique physical properties of sound will not only comprehend the fantastic capabilities of the current REMS technology but will also anticipate the incredible and yet-to-be discovered potential applications of acoustics science to the human body. This stuff is just amazing, and we are just starting to see the future of its use in all types of medical screening applications. The individuals who have chosen to use REMS are building solid bone health programs on a technology that is growing and developing!

While I am known in some circles to enjoy hating on DXA, I still am willing to give credit where credit is due. DXA is truly amazing technology. Period. The only reason why we are even writing this post and discussing osteoporosis these days is because of the advent of DXA. DXA was the first method with the necessary accuracy to determine BMD. I would stake the claim that all of the social media REMS-haters/DXA-lovers out there really don’t have the slightest clue about how complex DXA really is - the complex physics and mathematics that are behind every DXA scan and the true marvel of technology that it really is. The people who came up with DXA were obviously brilliant. As Bone Healthcare providers, we owe a lot to the developers of DXA and to those medical practitioners who initially started applying DXA to their Bone Health practices when it was a brand-new technology and not widely used or understood and not accepted by the general medical community. Those individuals have persisted and persevered and to this day they continue to promote the need for public awareness of Bone Health. DXA has been the foundation of their treatment of this crippling disease for over thirty years. And thirty years without change is a really long time! Maybe too long?

REMS is a completely different technology from DXA, based on a completely different manner in how the information on the properties of bone is collected (sound vs. ionizing radiation). Sound is a form of mechanical energy and therefore it can be modified by the physical properties of what it interacts with. REMS is an examination of bone with ultrasound - a single frequency soundwave is used to insonify the target bone (3.5MHz). The properties of the bone will alter the sound wave and the backscatter radio frequency wave (the echo) carries information on the physical properties of the bone that it just interacted with - just like the physical properties of a canyon will modify your voice when you yell into it and listen for your echo (see our second FB post). What that means is that a bone that has “good” density and “good” quality will alter the index sound wave differently than a bone with “bad” density and/or “bad” quality. REMS is then capable of analyzing the sound wave to generate the BMD value and Fragility Score. However, it is also important to realize that during this process, bone isn’t the only structure or tissue in our body that will interact with the sound wave and generate an echo. Therefore, the echo generated during a REMS study is an extremely complex soundwave and is actually composed of many different soundwaves coming from many different structures (bone, disc, cartilage, bone spurs, intestines, car keys). However, REMS isn’t confused by this complex soundwave. Utilizing a mathematical analysis (Fast Fourier Transformation) the complex sound wave is broken down into its component sound waves. The first level of filtering then analyzes the component soundwaves - artifacts that would be caused by non-bone sources are removed at this point because bone structure will have a specific spectral signature (fingerprint) that can be compared to a standard spectral database. Only the waveforms that meet the spectral requirements of bone are selected for the next level of analysis. If, for some reason, REMS was not able to detect and isolate enough sound waves that have the spectral criteria of bone then REMS does not proceed to the next step of analysis and requests that the examiner repeat the study. This eliminates the well known artifact error that plagues DXA scans (I couldn’t help myself).

In the case where there are a sufficient number of acceptable waveforms isolated during the scan, REMS will continue to process the information to generate an Osteoporosis Score (OS) and a Fragility score (FS). The Osteoporosis Score is derived from a sound wave that is significantly modified by bone density. This sound wave is then analyzed and is used to generate the OS. The OS is converted to a BMD value utilizing databases of bone ranging from normal to osteoporotic. The BMD value then is used to determine the T-score using the NHANES database. At this stage, limitations in REMS are due to the fact that all databases are based on the process of averaging a large group of individuals. Even though the process of REMS assessment of bone is not affected by the size of a bone that is being analyzed (soundwaves are not affected by bone size), the results of the REMS studies are constrained to the databases that are developed on population averages that tend to skew the results of individuals who may not be an “average” size. Unfortunately, DXA is also affected by the “averaging effects” of the available databases. However, the DXA issue is compounded by an intrinsic bone size error artifact due to the properties of x-ray and the method by which the bone is analyzed during a DXA scan (there I go again!). Back to REMS!

The FS is derived in a similar manner to the OS. A single sound wave that is affected by structural components of the bone is isolated and analyzed. It is then compared to a database of individuals who are fragile (have sustained a fragility-type fracture) or non-fragile (have not sustained a fragility-type fracture). The FS will then represent the probability of fracturing. The FS measurement has been shown in a recent study to be more sensitive and more specific in fracture prediction than either REMS or DXA BMD measured in the same study.

So, up to this point there really weren’t a lot of places where REMS could give a “wrong” result. If a study is determined by REMS to have insufficient data for analysis - it was found to be “noisy” REMS will automatically discard the study and there will be no data reported. Therefore, an incorrect report will not be generated! So where do the errors come from? Why then, were there a number of REMS assessments determined to be “erroneous” and therefore discarded from the large multi-center studies from Italy and the European Union that were published in the peer-reviewed literature and established that REMS was an equivalent method of bone densitometry as compared to DXA? Is this the REMS scandal that social media has been waiting for?

Well not really. In the largest multi-center study performed in the EU where 4307 patients were initially recruited, an average (hip and spine) 348 of the patients were excluded due to REMS errors - about 8.2%! Wow! Is that a lot? Probably not, because in the same study, an average 374 (hip and spine) patients were excluded due to DXA errors. About 8.8%! (the numbers were averaged in order to save space - this post is already long enough!). Although this may not be as dramatic as some may have hoped for, it actually does show that there can be “erroneous” REMS reports. Now, unfortunately this is the point where some of the other social media posts stopped and didn’t ask the next appropriate question - where do the REMS errors come from? But that's why we’re here! To make sure that all the facts are presented!

In general, errors can be made in the acquisition of any type of data. Obviously, that can also occur during a REMS scan. Every technology also has a learning-curve and if the individual doing the scan is not experienced, then errors in method and settings can be made. In REMS method errors will usually result in an ND (non-diagnostic) will not generate a report (unlike DXA - that will still provide a report). However, during REMS examinations, if appropriate technique is utilized when performing an examination but the settings are inappropriate, a report may be generated that is suboptimal (or just plain wrong). And in some cases those results will be reported. So where are the mistakes made? The Focal Setting (where REMS is supposed to focus the sound wave) and the Depth Setting (how deep is the bone that needs to be analyzed) are parameters that are chosen by the examiner. These parameters have optimal settings and if they are not optimized resulting data may not be correct. This tends to be a “newbie” mistake that is eliminated with experience. Noted in the above cited paper “ Concerning REMS errors, they were typically associated with wrong settings at acquisition parameters or with incomplete adherence to indications provided by the software and/or the user guide.” Also, several other papers recommended that REMS examiners performing research work have “at least 3 months of experience.”

It has been our experience, that experience solves the majority of REMS issues (not all, but the vast majority). Several articles also describe difficulty with obtaining satisfactory results with very large individuals (DXA has the same issue). This has been our experience and we have learned how to deal with not only issues of some folks being big, but also individuals who are very small. Now, REMS does not have an issue with bone size like DXA does - however, very small and thin individuals also present an examining challenge. We have learned through experience (and with Vicki’s help) that the maximum and minimum settings on the EchoS, if used during the spine examination, tend to give suboptimal or even incorrect results. How do we know that? If the spine results are very discordant from (do not match) the hip examinations that are performed, then there is something wrong (please refer to our previous post on how to understand REMS results). By re-performing the examination utilizing the next level of settings which often makes the examination more technically challenging to perform, the results will be much more concordant and more likely to truly reflect the true condition of the bone.

So, there are some challenges to performing a REMS assessment and it is important for someone performing a REMS to adhere to the basic principles while performing the examination. Pretty much, as long as the image is centered on the viewing screen and minimum or maximum settings are not used, the error rate of any REMS will match industry standards. The REMS report generated can be trusted to provide an accurate assessment of the condition of the bone being examined!

P.S. - one more REMS challenge. Gas! Although part of a healthy digestive tract, gas will make a REMS examination very difficult. That is why we perform our REMS in the mornings and ask our patients to skip breakfast. It’s not impossible to do the test after eating but it just makes the test a little harder to do!

Hopefully, this post has answered some of your additional questions about REMS. And, as much as I had managed to restrain myself and not hate-on DXA, logically, the next post will have to discuss errors associated with DXA. All kidding aside, the intent is to provide a little more of an explanation behind some of the errors that are notorious to DXA and that we love to point out. With the intent to help anyone who had a DXA why the DXA results may not be considered to be correct.