Background/Purpose: Children with Systemic Lupus Erythematosus (SLE) are often understudied compared with adults. Using the electronic health record (EHR) can increase sample size and the diversity of SLE patients while providing longitudinal data on comorbidities. We assessed differences in comorbidities in pediatric vs. adult-onset SLE using an EHR-based phenome-wide association study (PheWAS). PheWAS is a systematic and efficient approach that compares two groups using ICD-9 codes.

Methods: We used our validated algorithm of ≥ 4 counts of the SLE ICD-9 code (710.0) and ANA positive > 1:160 while excluding dermatomyositis and systemic sclerosis ICD-9 codes to identify SLE cases in a de-identified EHR called the Synthetic Derivative (SD). The SD contains over 2.5 million subjects with longitudinal clinical data. Our algorithm has an internally validated positive predictive value of 94% and a sensitivity of 86%. Pediatric-onset SLE was defined as ≤ 18 years at first use of the SLE ICD-9 code vs. adult-onset ≥ 19. PheWAS was performed in pediatric vs. adult-onset adjusting for sex and race in logistic regression models and correcting for multiple testing using Bonferroni. PheWAS excludes subjects that have a one time count for a code to minimize the effect of coding errors.

Results: We identified 93 pediatric-onset and 1005 adult-onset patients with SLE. Pediatric-onset patients had a mean age at first SLE code of 15 ± 3 years versus adult-onset at 43 ±15. Both pediatric and adult-onset patients were predominantly female (87% vs. 90%, p = 0.34). Pediatric-onset patients were more likely African Americans (49% African American, 40% Caucasian, 8% Asian, and 3% Hispanic) vs. adult-onset (72% Caucasian, 24% African American, 3% Hispanic, and 1% Asian), (p < 0.001). Adjusting for sex and race, pediatric-onset patients had 13 codes that met the Bonferroni threshold for significance (p < 1.26 x 10^-4) including mostly nephritis codes (Table 1, Figure 1). Compared to adult-onset, pediatric-onset patients had more codes related to other SLE ACR criteria such as pleurisy/pleural effusion, pericarditis, thrombocytopenia, aplastic anemia, pancytopenia, convulsions, and non-infectious encephalitis (all p < 0.05).

Conclusion: Using a large EHR, pediatric-onset patients had an increased SLE disease burden with more ICD-9 codes related to ACR SLE criteria, particularly renal disease. These findings demonstrate the ability of PheWAS to replicate epidemiologic associations within subgroups of SLE patients and to function as an efficient discovery tool in the EHR.