Key points are not available for this paper at this time.
ABSTRACT: Background: Degenerative rotator cuff tears (DCTs) are the leading cause of shoulder pain, affecting 30% to 50% of individuals over 50. Current phenotyping strategies for DCT rely on heterogeneous combinations of procedural and diagnostic codes, leading to concerns about misclassification. We aimed to create universal phenotypic algorithms to classify DCT status across EHR systems. Methods: Using Vanderbilt University Medical Center's de–identified EHR system, we developed and validated two sets of algorithms – one requiring imaging verification and one without imaging verification – to identify DCT cases and controls. The algorithms used combinations of ICD and CPT codes, as well as natural language processing (NLP), to increase diagnostic certainty. Manual chart reviews by trained personnel blinded to case–control determinations were conducted to compute positive (PPV) and negative predictive values (NPV). Results: The algorithm development process resulted in five phenotypes with an overall predictive value of 94.5%. By approach: 1) code only cases that required imaging confirmation (PPV = 89%), 2) code only cases that did not require imaging verification (PPV = 92%), 3) NLP–based cases that did not require imaging verification (PPV = 89%), 4) code–based controls that were confirmed by imaging (NPV = 90%), and 5) code and NLP–based controls that did not require imaging verification (NPV = 100%). External validation demonstrated 94% sensitivity and 75% specificity for the code–only algorithms. The addition of NLP increased the number of identified cases without compromising predictive values. Conclusions: These algorithms represent an improvement over current phenotyping strategies and allows for EHR studies of unprecedented size
Herzberg et al. (Mon,) studied this question.