Autoimmune rheumatologic diseases
Systemic Lupus Erythematosus (SLE) is a systemic autoimmune disease with diverse manifestations. Electronic health records (EHR) are a rich source of information that can be used to understand the presentation of SLE. We assessed three SLE clinical classification criteria as a foundation for phenotype-based detection of SLE patients in EHR data. Performance was evaluated over 600 medical records from the Northwestern Medicine EHR system, 472 with definite SLE and 128 without, based on chart review. We developed three algorithms, based on the American College of Rheumatology (ACR), Systemic Lupus International Collaborating Clinics (SLICC) and proposed European League Against Rheumatism/American College of Rheumatology (EULAR/ACR) classification criteria using only structured data elements (diagnosis codes and lab results) to determine whether patients met the classification criteria for SLE. The SLE identification rate ranged from 49-73% across the algorithms; all had >=97% specificity and >=99% positive predictive value (PPV). Sensitivity ranged from 49-73% and negative predictive value (NPV) from 43-58%. The SLICC-based algorithm performed best, detecting 73% of patients with SLE, with 99% PPV, 73% sensitivity, 97% specificity and 58% NPV. All 3 algorithms detect a significant proportion of patients with SLE, with high PPV and specificity. Low NPV of the algorithms likely reflects undetected cases of SLE resulting from low detection of clinical and laboratory criteria that are not consistently documented in structured data. The algorithms may improve through use of natural language processing of physician notes for criteria that were difficult to detect using diagnosis codes and labs.