Rotator Cuff Tear

A patient sent me a message yesterday regarding her shoulder. She had slept poorly and woke up with some shoulder discomfort. After seeing her doctor, he referred her for a scan. Her MRI shows a tear in the supraspinatus muscle (part of the rotator cuff) and some osteoarthritic changes in the joint. She was worried about the tear. It got scarier when her friends told her that if she did not "fix it" the tear would become worse and her shoulder would wear out.

I told her about a recently published study done where 602 participants had  scans done on both shoulders. (Ibounig et al, 2026). Median age of the participants was 59 (range from 41-76 years old) and 52 percent were female. 

Rotator cuff abnormalities were found in 98.7 percent of painful shoulders. There were 25 percent with tendinopathy, 62 percent partial thickness tears and 11 percent with full thickness tears (FTT).

More importantly, rotator cuff (RC) abnormalities were seen in 96 percent of pain free shoulders. Full thickness tears of the RC were in 7 percent of shoulders that did not hurt. 

The prevalence and severity of abnormalities increased with age, but did not differ between sexes. 

Only FTT were more prevalent in painful shoulders (14.6 percent) than in non-painful shoulders (6.5 percent). However this difference decreased after "adjustment" (absolute difference, 0.8%; 95% CI, -3.4% - 6%). This means that once the researchers accounted for age, the apparent link between having shoulder pain and a FTT vanished.

At first glance, FTT looked more common in painful shoulders (14.6%) versus non-painful shoulders (6.5%). However, older people are more likely to have shoulder pain and a FTT just from normal aging.

"Adjustment" is a statistical technique that levels the playing field as it allows reserachers to compare painful and non-painful shoulders of the exact same age. When they did this, the true difference was just 0.8 percent. This is a very tiny margin, meaning the presence of a tear does not reliably explain why a shoulder hurts.

95% CI, -3.4%-6%. When the confidence interval (CI) crosses zero (goes from a negative to positive number), it means there is no statistically significant difference. The negative side means it is statistically possible that FTT are actually less common in painful shoulders by 3.4 percent.

The positive side means it is statistically possible FTT are more common in painful shoulders by 6 percent.

The math proves that one cannot look at a FTT on an MRI an automatically assume it is the cause of someone's shoulder pain. A person with a painful shoulder is statistically no more likely to have a full thickness tear than a person of the same age with completely pain free shoulders.

I told my patient that her 'abnormalities' on the scan were actually just aging. What matters most for her (and others) is function. Is there weakness and/ or pain when she moves her shoulder?

Can she sleep without pain and can she play pickleball without pain? If all her answers are no, she most likely will NOT need an operation. Just because abnormalities are seen on MRI does not mean it can cause pain or needs to be fixed.

Reference

Iboinig T, Jarvinen TLN, Raatikainen S et al (2026). INcidental Rotator Cuff Abnormalities ON Magnetic Resonance Imaging. JAMA Intern Med. 186(4): 406-414. DOI: jamainternmed..2025.7903