Background: Performing adequately powered clinical trials in pediatric diseases such as systemic lupus erythematosus (SLE) is challenging. Improved recruitment strategies are needed for identifying patients.
Design, Setting, Participants, and Measurements: Electronic health record (EHR) algorithms were developed and tested to identify children with SLE both with and without lupus nephritis. We used single center EHR data to develop computable phenotypes comprised of diagnosis, medication, procedure, and utilization codes. These were evaluated iteratively against a manually assembled SLE patient database. The highest performing phenotypes were then evaluated across institutions in PEDSnet, a national healthcare systems network of >6.7 million children. Reviewers blinded to case status used standardized forms to review random samples of cases (n=350) and non-cases (n=350).
Results: Final algorithms consisted of both utilization and diagnostic criteria. For both, utilization criteria included ≥2 in-person visits with nephrology or rheumatology and ≥60 days follow-up. SLE diagnostic criteria included absence of neonatal lupus, ≥1 hydroxychloroquine exposure, and either ≥3 qualifying diagnosis codes separated by ≥30 days, or ≥1 diagnosis code and ≥1 kidney biopsy procedure code. Sensitivity was 100% (95%CI, 99-100); specificity, 92% (95% CI, 88-94); positive predictive value, 91% (95%CI, 87-94); negative predictive value, 100% (95%CI, 99-100).
Lupus nephritis diagnostic criteria included either ≥3 qualifying lupus nephritis diagnosis codes (or SLE codes on same day as glomerular/kidney codes) separated by ≥30 days, or ≥1 SLE diagnosis code and ≥1 kidney biopsy procedure code. Sensitivity was 90% (95%CI, 85-94); specificity, 93% (95% CI, 89-97); positive predictive value, 94% (95%CI, 89-97); negative predictive value, 90% (95%CI, 84-94).
Algorithms identified 1,508 children with SLE at PEDSnet institutions (537 with LN), 809 of whom were seen in the past 12 months.
Conclusions: EHR-based algorithms for SLE and Lupus nephritis demonstrated excellent classification accuracy across PEDSnet institutions.