As I’ve blogged before, patients and physicians often view MRI results like a report from an Oracle, in that they believe it will elucidate all of life’s mysteries. However, the scientific reality is that MRIs are pretty horrible at diagnosing the cause of pain. Now a new study shows, not surprisingly, that shoulder MRI results are equally really bad at accurately diagnosing the cause of shoulder pain.
I think it’s patient frustration and the comfort that comes from certainty (even if that certainty is actually artificial) that causes patients to focus on the MRI report. Like many physicians, I am also a patient in certain body areas and I know I have the same tendency to want certainty. Having said that, research study after research study continues to show that your MRI is no better than chance at figuring out why you hurt. Take for example knee MRIs for meniscus tears. An entire multi-billion dollar industry was created to surgically resect these tears after MRI scans became popular in the 80s and this industry is alive and well today. However, the world’s largest and best study on meniscus tears shows that patients with and without knee pain have the same likelihood of having a meniscus tear on an MRI scan. This could be why other recent research shows that surgically treating these meniscus tear phantoms is no more effective than a placebo surgery or simple physical therapy exercises. The same holds tue for many other body areas where operating on an MRI finding has become the norm including hip labrum tears, hip impingement, and low back spinal stenosis. How about shoulder MRI? Surely that must fare better as countless surgeries are performed annually on rotator cuff and labral tears.
In one recent study that was published a few months ago, two board certified musculoskeletal radiologists sought to find out if shoulder MRI was accurate in predicting who had shoulder pain. They took 30 older patients and divided them into three groups-those with no history of or active shoulder pain, those who had a previous history of shoulder pain but didn’t have it now, or those with active shoulder pain. They then both read the shoulder x-rays and MRIs from these patients while remaining blind to which patient was in each group (i.e. they didn’t know if the MRI belonged to someone who had never had shoulder pain, used to have pain or had pain). They then read these films using a structured approach so they wouldn’t miss anything. The results? On X-ray, there was no significant difference between the three groups in terms of arthritis (meaning the patients without pain and with shoulder pain all had equal amounts of the disease). Pissed off shoulder tendons and tears of the rotator cuff were present in the majority in each group, again with no differences in the 20 patients with a history of or who had shoulder pain and the pain free group. Finally, while labral abnormalities were rare among all groups, the shoulder pain groups had no more than the pain free group.
The upshot? If this were the first study of it’s kind to show that MRI findings are common in pain free individuals, I’d be surprised. However, there are now more than 100 studies just like this for pretty much all body parts that might possibly hurt-hence these results fit with other studies that we have. Is MRI useless? No, it can be a very helpful technology when used very carefully with a good physical exam and as a piece of a very large puzzle of data amassed on what might be causing the patient’s pain. Why then do doctor’s fawn over these images and schedule surgeries based on the findings and a cursory exam? That is in fact the multi-billion dollar question-as the research continues to show that this approach is a recipe for unnecessary surgeries, Why then do all of these NFL guys get surgery based on an MRI image? First, that’s a bit different. These guys are all young, so it’s less likely they have wear and tear age related changes that might be misinterpreted as a major MRI abnormality. Second, they just had a specific trauma to a body part and that part gets imaged that day or the next. The injured area is still swollen and that can be seen on MRI. The average American gets imaged months to years after the original injury, if there was one. Hence, your injury likely isn’t like the NFL guy’s you saw carted off the field!