Search

Large collaborative research networks provide opportunities to jointly analyze multicenter electronic health record (EHR) data, which can improve the sample size, diversity of the study population, and generalizability of the results. However, there are challenges to analyzing multicenter EHR data including privacy protection, large-scale computation resource requirements, heterogeneity across sites, and correlated observations. In this paper, we propose a federated algorithm for generalized linear mixed models (Fed-GLMM), which can flexibly model multicenter longitudinal or correlated data while accounting for site-level heterogeneity. Fed-GLMM can be applied to both federated and centralized research networks to enable privacy-preserving data integration and improve computational efficiency. By communicating a limited amount of summary statistics, Fed-GLMM can achieve nearly identical results as the gold-standard method where the GLMM is directly fitted to the pooled dataset. We demonstrate the performance of Fed-GLMM in numerical experiments and an application to longitudinal EHR data from multiple healthcare facilities.

Amebiasis is an endemic disease and a public health problem throughout Mexico, although the incidence rates of amebic liver abscess (ALA) vary among the geographic regions of the country. Notably, incidence rates are high in the northwestern states (especially Sonora with a rate of 12.57/100,000 inhabitants) compared with the central region (Mexico City with a rate of 0.69/100,000 inhabitants). These data may be related to host genetic factors that are partially responsible for resistance or susceptibility. Therefore, we studied the association of the HLA-DRB1 and HLA-DQB1 alleles with resistance or susceptibility to ALA in two Mexican populations, one each from Mexico City and Sonora. Ninety ALA patients were clinically diagnosed by serology and sonography. Genomic DNA was extracted from peripheral blood mononuclear cells. To establish the genetic identity of both populations, 15 short tandem repeats (STRs) were analyzed with multiplexed PCR, and the allelic frequencies of HLA were studied by PCR-SSO using LUMINEX technology. The allele frequencies obtained were compared to an ethnically matched healthy control group (146 individuals). We observed that both affected populations differed genetically from the control group. We also found interesting trends in the population from Mexico City. HLA-DQB1*02 allele frequencies were higher in ALA patients compared to the control group (0.127 vs 0.047; p= 0.01; pc= NS; OR= 2.9, 95% CI= 1.09-8.3). The less frequent alleles in ALA patients were HLA-DRB1*08 (0.118 vs 0.238 in controls; p= 0.01; pc= NS; OR= 0.42, 95% CI= 0.19-0.87) and HLA-DQB1*04 (0.109 vs 0.214; p= 0.02; pc= NS; OR= 0.40, 95% CI= 0.20-0.94). The haplotype HLA-DRB1*08/-DQB1*04 also demonstrated a protective trend against the development of this disease (0.081 vs. 0.178; p=0.02; pc=NS; OR= 0.40, 95% CI= 0.16-0.93). These trends suggest that the prevalent alleles in the population of Mexico City may be associated with protection against the development of ALA.