Development and Validation of an Electronic Medical Record Algorithm to Identify Phenotypes of Rotator Cuff Tear.
Chan Gao, Run Fan, Gregory D Ayers and 5 others
PMID 32198840WHAT IT FOUND
An electronic record algorithm identified rotator cuff tear with sensitivity 0.68 and specificity 0.89 in validation, but it was better at ruling in than ruling out and was not tested outside one center.
Key findings
01In validation, the algorithm had sensitivity 0.68 and specificity 0.89 for rotator cuff tear.
02The final clinical model used age, sex, CPT repair codes, physician notes, and radiology reports, with CPT, physician notes, and radiology reports reported as significant predictors.
03In the development chart review, 193 true rotator cuff tear and 100 true normal rotator cuff patients were confirmed, while 707 subjects could not be definitively classified.
STILL TO COME
How it was doneWhat they foundWhat it means for PTs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The algorithm was developed and validated only within one medical center, so its performance in other EMR systems or patient populations is unknown. Many randomly selected patients could not be classified because imaging or operative information was missing: 707 subjects in the development review. Sensitivity was 0.68 and specificity was 0.89, so the algorithm was better at ruling in rotator cuff tear than ruling it out. Patients without MRI or operative documentation were excluded, so less severe cases may not be represented. The algorithm did not distinguish traumatic from degenerative rotator cuff tears. ICD-10 codes were not included because too little data were available. Normal cuff status was confirmed on only one shoulder, and the status of the other shoulder was unknown.
Declared interests
The paper lists NIH extramural research support. No other funding or conflict declaration is given in the supplied text.
The easy way to misread this
Do not use this algorithm to decide whether a patient has a rotator cuff tear. It had sensitivity 0.68 in validation, and it was not tested outside one center.