Lossless integration of multiple electronic health records for identifying pleiotropy using summary statistics

• DOI: 10.1038/s41467-020-20211-2
• Publication Date: 2021-01-08T11:13:01Z
• Authors: Ruowang Li; Rui Duan; Xinyuan Zhang; Thomas Lumley; Sarah Pendergrass; Christopher Bauer; Hakon Hakonarson; David S. Carrell; Jordan W. Smoller; Wei-Qi Wei; Robert Carroll; Digna R. Velez Ed...; Georgia Wiesner; Patrick Sleiman; Josh C. Denny; Jonathan D. Mosley; Marylyn D. Ritchie; Yong Chen; Jason H. Moore
• Keywords: Pleiotropy

Increasingly, clinical phenotypes with matched genetic data from bio-bank linked electronic health records (EHRs) have been used for pleiotropy analyses. Thus far, analysis using individual-level EHR has limited to one site. However, it is desirable integrate multiple sites improve the detection power and generalizability of results. Due privacy concerns, patients' are not easily shared across institutions. As a result, we introduce Sum-Share, method designed efficiently perform analysis. Sum-Share requires only summary-level round communication each site, yet produces identical test statistics compared that pooled data. Consequently, can achieve lossless integration datasets. Using real eMERGE, able identify 1734 potential pleiotropic SNPs five cardiovascular diseases.