top of page

Peering Beneath The Surface Might Not Be As Insightful As We Once Thought

Shoulder pain is a common musculoskeletal complaint among athletes and the general population. Many individuals seeking consultation undergo advanced imaging such as Magnetic Resonance Imaging (MRI) to identify underlying pathoanatomical issues. However, evidence continues to illustrate the faulty premise behind this approach, as we previously discussed in our guide to shoulder pain.

Often, the same findings on MRI can be found in asymptomatic populations or have little to no correlation to symptom development and/or persistence [2-4]. Incidentaloma is a word that has recently emerged in other areas of healthcare and is defined as - “...incidentally found asymptomatic tumor” [5]. Since the inception of this term, many clinicians (myself included) use it to describe any incidental imaging findings (i.e., not just tumors) that are unrelated to the original diagnostic inquiry. Often incidentalomas cause unnecessary worry by both the clinician and patient, leading to further evaluation and treatment resulting in increased healthcare utilization, complications, and costs.

Dr. Richard Hayward, a pediatric neurosurgeon, developed an acronym to describe this healthcare experience: VOMIT – victims of modern imaging technologies [6]. As observed in the context of low back pain, we are increasingly questioning the significance of findings on shoulder imaging with respect to patient


We already have data demonstrating

many supposed “issues” are readily

observed in the asymptomatic aging

population. For example, a

2014 systematic review

on the prevalence of rotator cuff disease found the prevalence of abnormalities increased

from 9.7% in 20 year olds to 62% in

80 year olds and older. Furthermore,

the prevalence increased similarly regardless of symptoms or shoulder dislocation. The authors conclude:

“The prevalence of rotator cuff abnormalities in asymptomatic people is high enough for degeneration of the rotator cuff to be considered a common aspect of normal human aging and to make it difficult to determine when an abnormality is new (e.g., after a dislocation) or is the cause of symptoms." [2]

We also have data that MRI abnormalities are identified early in life.

Pennock et al recruited 23 asymptomatic male little league baseball players (average age 11.4 years) and performed bilateral shoulder MRIs. The MRIs revealed 17 asymmetric “abnormalities” in dominant shoulders of 12 players

(52%). 4 out of the 12 players had multiple abnormalities. The non-dominant arm MRIs had 2 abnormalities (labral tears). Overall, there was a 6-fold higher rate of “abnormalities” in the throwing arm of the athletes. The most common abnormalities


  • Edema/widening of proximal humeral physis (5 players)

  • Labral tear (4 players)

  • Partial rotator thickness tear (4 players)

  • Hypertrophy of acromioclavicular joint capsule (2 players)

  • Subacromial bursitis (1 player)

  • Cystic change of greater tuberosity (1 player) [7]

This raises the question of whether these “issues” we observe imaging in athletic

populations is actually the result of normative sport-specific adaptations. We have

similar findings from other studies:

This month’s research review article seeks to provide further insight into MRI findings as they relate to shoulder symptoms.


The authors’ primary purpose was to identify and compare the frequency of MRI tissue abnormalities in both shoulders (symptomatic and asymptomatic) of participants presenting with unilateral shoulder pain. Their secondary objective was to compare the interrater agreement of MRI-identified tissue abnormalities between a radiologist and shoulder orthopedic surgeon. The authors’ hypotheses were:

  1. “Both shoulders would present a high prevalence of alterations, and

  2. substantial agreement would occur between the shoulder surgeon and radiologist.” [1]


Participants from the local community were recruited via advertisements online and printed flyers posted. Inclusion criteria consisted of the following:

  • Self-reported unilateral atraumatic shoulder pain

  • Pain reported for a minimum of four weeks since initial onset

  • Pain symptoms made worse with activity

  • Ability to fully and actively elevate arm (according to the authors, ~150 degrees or greater as measured by digital inclinometer)

In total, 347 individuals were screened for inclusion. 224 participants were ultimately excluded for the following reasons (n):

  • Appointment missed (53)

  • Neck-related pain (43)

  • Upper trapezius muscle complaints (32)

  • Bilateral shoulder pain (31)

  • Fracture or trauma (25)

  • Recurrent glenohumeral dislocation (19)

  • Adhesive capsulitis (12)

  • Fear of MRI (6)

  • Receiving physical therapy (2)

  • MRI contraindicated (1)

In total, 123 participants were included (246 shoulders). The participants reported persistent but intermittent unilateral shoulder pain made worse with activity. Ongoing symptoms were reported for an average of 35.8 months since initial onset (standard deviation, s.d., 58.5 months, range 1 - 360 months). Average age was 39.4 years (s.d., 15.23 years, range of 18-77 years). 66 participants were men and 57 women.

All participants received non-contrast bilateral shoulder MRIs (i.e., both

symptomatic and asymptomatic sides) using a Magnetom Essenza MRI machine

(1.5 T). Images were obtained in the sagittal, coronal, and axial planes with participants in a supine position with arms by their side. The MRIs were then anonymized and randomized for independent review by an orthopedic surgeon and a radiologist. The radiologist examined the images based on his standard clinical practice and the orthopedic surgeon utilized an online structured form. See the chart (right) for a list of diagnostic terms that were utilized. SPSS Statistics was used to analyze the data. Findings were compared between symptomatic and asymptomatic shoulders via the Yates' chi-square test (χ2) or the Fisher exact test when expected count was less than 5; a P -value of less than 0.05 was set for statistical significance. Rater agreement was assessed by total observed agreement and Coen’s K index. The total observed agreement was calculated as:

  • Sum of simultaneous positive and negative classifications / Number of total observations

Cohen’s K Index was used to assess inter-rater agreement between radiologist and

orthopedic surgeon. The level of agreement between raters was classified as follows:

  • No agreement = 0

  • Slight = 0.01 - 0.20

  • Fair = 0.21 - 0.40

  • Moderate = 0.41 - 0.60

  • Substantial = 0.61 - 0.80

  • Almost Perfect = 0.81 - 1.00


The authors reported:

“MRI alterations were highly observed on all scans, and similar prevalence data were generally noted in both symptomatic and asymptomatic shoulders (Tables I and II). [emphasis mine]

Additionally, rotator cuff tendinopathy and AC joint alterations were highly prevalent in both shoulders (according to radiologist ~90% and orthopedic surgeon ~75%). See table 1 (below) for additional findings. According to the radiologist’s readings, there were no significant differences in prevalence of MRI findings between symptomatic and asymptomatic shoulders, except for partial-thickness infraspinatus tears. The shoulder surgeon reported a higher prevalence of full-thickness tears of the supraspinatus tendon and glenohumeral osteoarthritis in symptomatic vs asymptomatic shoulders (see table 2 in the published

article). The authors state,

“Overall, only the prevalence of full-thickness tears and glenohumeral OA presented statistically significant differences between the symptomatic and asymptomatic sides.”

They go on to state that rotator cuff tear size, location, and retraction may play a role in the risk of subsequent symptom development and functional decline. It is important to recall that statistical significance does not equate to clinical significance. The authors also state:

“Our investigation questions the clinical utility of MRI for treatment planning because a high prevalence of alterations in asymptomatic shoulders was observed. This finding indicates that although MRI can accurately identify tissue pathology, it cannot discriminate whether that altered anatomy is associated with specific clinical findings. This is of particular importance to physicians, surgeons, and physical therapists who might use the MRI findings to guide the decision toward a treatment plan.”

It’s also worth noting that the shoulder surgeon identified more glenohumeral osteoarthritis than the radiologist (radiologist = 2 symptomatic, 1 asymptomatic and shoulder surgeon = 13 symptomatic, 4 asymptomatic). Interestingly, the total observed agreement for MRI findings between the radiologist and shoulder surgeon ranged from 44.71% - 98.14%; with best agreement for supraspinatus atrophy, glenohumeral osteoarthritis, and long head of biceps tendon alteration at ~90%. Kappa Indexes for agreement ranged from 0.00 - 0.51, with the best k-index for partial-thickness tears at 0.38, long head of biceps tendon alterations at 0.44, and cysts in humeral tuberosities at 0.51.

Kappa indexes represent the level of agreement between raters for the same finding. At best, the raters demonstrated moderate agreement and at worst, didn’t agree at all on a particular finding. Kappa indexes such as this should prompt serious consideration of the inability to replicate a finding; replicability is a fundamental aspect of the scientific method. The lack of replicability between raters becomes more concerning when treatment selection is being hinged to such findings.

Why does this article matter?

Currently in our healthcare system, often imaging is ordered in hopes of elucidating why a patient has developed symptoms by identifying a pathoanatomical issue (i.e., not textbook “norm”). However, in the musculoskeletal world this is proving to be more and more difficult. We continue to have mounting evidence demonstrating such findings in asymptomatic populations or little to no correlation to symptom development and persistence. This article continues to build on prior evidence as it relates to the shoulder. The authors state:

“The results of this study demonstrated a high prevalence of bilateral MRI alterations in individuals with unilateral shoulder pain, confirming our hypothesis. Few differences between symptomatic and asymptomatic shoulders were observed.”

Although a linear-cause effect model is advised against for most pain based issues, if we wish to follow such thinking then we need to be able to identify some level of correlation to imaging findings and symptoms, more importantly - these findings directly influence patient management. Based on the findings in this article, it is difficult to say a pathoanatomical finding is highly correlated to symptoms given similar findings were found in both shoulders (i.e., both asymptomatic and symptomatic). One may wish to argue that perhaps these findings are a contributor to the development of future symptoms, but again - the evidence doesn’t appear supportive at this time nor is it necessary to alter management for the majority of findings [2-4]. Regardless, we must remember we are treating the person and not just a biological silo.

A secondary finding of this article is the lack of agreement between the radiologist and shoulder surgeon. The kappa index demonstrates no to moderate agreement (0.51) based on the authors’ rating system (see above). This level of interrater agreement further demonstrates the overall subjectivity of MRI interpretation. Reviewing Appendix A demonstrates how the labels provided for issues such as rotator cuff tears, AC joint issues, glenohumeral osteoarthritis, musculotendinous retraction, and adhesive capsulitis have an inherent subjectivity to defining the terms which leaves interpretation open to the observer. We’ve seen similar issues with MRI interpretations in other studies. Recall Herzog 2017 where a single individual visited 10 MRI centers for a lumbar MRI, and the authors found:

“... 49 distinct findings reported related to the presence of a distinct pathology at a specific motion segment. Zero interpretive findings were reported in all 10 study examinations and only one finding was reported in nine out of 10 study examinations. [emphasis mine]

They concluded:

”This study found marked variability in the reported interpretive findings and a high prevalence of interpretive errors in radiologists' reports of an MRI examination of the lumbar spine performed on the same patient at 10 different MRI centers over a short time period. As a result, the authors conclude that where a patient obtains his or her MRI examination and which radiologist interprets the examination may have a direct impact on radiological diagnosis, subsequent choice of treatment, and clinical outcome.”

Recall the acronym VOMIT (Victims of Modern Imaging Technologies) describing the healthcare experience were unnecessary imaging leads to increased healthcare utilization and cost. We have good evidence on this in the context of low back pain. For example, Lemmers 2019 completed a systematic review examining imaging versus no imaging for low back pain and subsequent healthcare utilization, cost, and absenteeism from work. The authors concluded,

“This study concludes that imaging in patients with low back pain does increase costs and healthcare utilization. There are indications that it also leads to higher absence from work. This is unwarranted for both patients and society since we know that imaging in low back pain has no health benefit.”

One example of this issue as it relates to the shoulder is subacromial impingement syndrome. This is a common narrative provided to patients and often includes imaging findings such as loss of acromial space, acromion morphology, rotator cuff tears/tendinopathy, etc. Cuff 2017 demonstrated how this narrative is damaging to patient beliefs, behaviors, and future treatment selections. The authors interviewed 9 patients with the diagnosis of subacromial impingement syndrome referred for physiotherapy. The authors argue that the diagnosis of “Subacromial Impingement Syndrome” reflects a biomedical narrative that instills fear and avoidance of movement among patients. Patients are told they have a bony outgrowth that “impinges” with particular shoulder movements, thus damaging, degrading, or “fraying” underlying tissues over time. In response, many naturally become fearful and, potentially unbeknownst to them, begin self-selecting movements they deem “acceptable” vs “unacceptable” for fear of precipitating further shoulder damage. As one can imagine, this can become quite detrimental to daily function and desired activity.

There are emerging arguments to alter radiology reports or include epidemiology for imaging findings. Bossen 2013 provided insight to this topic with their study where MRI reports were rewritten to an 8th grade reading level with neutral language and optimistic interpretations of radiology findings. The authors discuss their approach by altering word usage for a finding of tear :

“We used neutral descriptive words and the most optimistic interpretation based on current best evidence. For instance, words such as ‘‘tear’’ were replaced by more descriptive and accurate words such as hole, signal change, or defect. We also used analogies (eg, gray hair, bald spot) where appropriate.”

The authors found:

“Emotional response, satisfaction, usefulness, and understanding were all superior in MRI reports reworded for lower reading level and optimal emotional content and optimism. Given that patients increasingly have access to their medical records and diagnostic reports, attention to health literacy and psychologic aspects of the report may help optimize health and patient satisfaction.”

With all this said, we must still question whether imaging is even warranted in many shoulder case situations, but perhaps at least altering radiology reports may be a step in the right direction.

Finally, it’s worth mentioning a limitation of the Barreto study. Given the cross-sectional design of the study, it’s difficult to draw conclusions regarding the correlation for future development of symptoms (risk factors) and a better study to make stronger claims to this discussion would be prospectively designed where participants undergo an initial MRI and then are followed over a period of time (months - years) and potentially undergo follow-up MRIs to assess imaging alterations and symptom occurrences.

To close, the authors state -

“Our investigation questions the clinical utility of MRI for treatment planning because a high prevalence of alterations in asymptomatic shoulders was observed. This finding indicates that although MRI can accurately identify tissue pathology, it cannot discriminate whether that altered anatomy is associated with specific clinical findings.”

I’m in agreement with the above statement. This article further supports our need as clinicians to treat the person in front of us and not just “a shoulder case” or a particular finding on imaging. Anecdotally, many may have similar imaging findings and are pain-free and fully functioning at their desired level while others may have less “severe” findings and aren’t coping well, unable to function at a desired level, and continue to deal with persistent pain based symptoms. Hopefully such evidence will continue to demonstrate that pain is a multifactorial, individualistic experience and trying to reduce the person’s experience to imaging findings is too reductionist and overall unhelpful.

Appendix A:


  1. Barreto RPG, Braman JP, Ludewig PM, Ribeiro LP, Camargo PR. Bilateral magnetic resonance imaging findings in individuals with unilateral shoulder pain. Journal of shoulder and elbow surgery. 2019.

  2. Teunis T, Lubberts B, Reilly BT, Ring D. A systematic review and pooled analysis of the prevalence of rotator cuff disease with increasing age. Journal of shoulder and elbow surgery. 2014; 23(12):1913-1921.

  3. Tran G, Cowling P, Smith T, et al. What Imaging-Detected Pathologies Are Associated With Shoulder Symptoms and Their Persistence? A Systematic Literature Review. Arthritis care & research. 2018; 70(8):1169-1184.

  4. Connor PM, Banks DM, Tyson AB, Coumas JS, D'Alessandro DF. Magnetic resonance imaging of the asymptomatic shoulder of overhead athletes: a 5-year follow-up study. The American journal of sports medicine. ; 31(5):724-7.

  5. Chojniak R. Incidentalomas: managing risks. Radiologia brasileira. ; 48(4):IX-X.

  6. Hayward R. VOMIT (victims of modern imaging technology)—an acronym for our times. BMJ .2003;326(7401):1273.

  7. Pennock AT, Dwek J, Levy E, et al. Shoulder MRI Abnormalities in Asymptomatic Little League Baseball Players. Orthopaedic journal of sports medicine. 2018; 6(2):2325967118756825.

  8. Del Grande F, Aro M, Jalali Farahani S, Cosgarea A, Wilckens J, Carrino JA. High-Resolution 3-T Magnetic Resonance Imaging of the Shoulder in Nonsymptomatic Professional Baseball Pitcher Draft Picks. Journal of computer assisted tomography. ; 40(1):118-25.

  9. Lee CS, Stetson WB, Goldhaber NH, Davis SM, Brock A, Wosmek J. Magnetic Resonance Imaging Findings in Asymptomatic Elite Volleyball Players Arthroscopy: The Journal of Arthroscopic & Related Surgery. 2017; 33(10):e58-e59. Johansson F, DeBri E, Swärdh L, et al. MRI FINDINGS IN THE SHOULDER OF COMPLETELY ASYMPTOMATIC ADOLESCENT ELITE TENNIS PLAYERS Br J Sports Med. 2014; 48(7):612.2-612.

  10. Fredericson M, Ho C, Waite B, et al. Magnetic resonance imaging abnormalities in the shoulder and wrist joints of asymptomatic elite athletes. PM & R : the journal of injury, function, and rehabilitation. 2009; 1(2):107-16.

  11. Herzog R, Elgort DR, Flanders AE, Moley PJ. Variability in diagnostic error rates of 10 MRI centers performing lumbar spine MRI examinations on the same patient within a 3-week period. The spine journal : official journal of the North American Spine Society. 2017; 17(4):554-561.

  12. Lemmers GPG, van Lankveld W, Westert GP, van der Wees PJ, Staal JB. Imaging versus no imaging for low back pain: a systematic review, measuring costs, healthcare utilization and absence from work Eur Spine J. 2019; 28(5):937-950.

  13. Cuff A, Littlewood C. Subacromial impingement syndrome - What does this mean to and for the patient? A qualitative study. Musculoskeletal science & practice. 2018; 33:24-28.

  14. Bossen JK, Hageman MG, King JD, Ring DC. Does rewording MRI reports improve patient understanding and emotional response to a clinical report? Clinical orthopaedics and related research. 2013; 471(11):3637-44.

70 views0 comments

Recent Posts

See All
bottom of page