Table 3 Best practices being developed by the eMERGE sites

Marshfield

Enhance internal EMRs to capture data in a structured format. This may involve changing existing input points in the record. Information validated against questionnaire data where applicable; Develop and evaluate a computer-based consenting process along with revisions to the current written informed consent document for our general biobank; Development of validated electronic algorithms for cataract, HDL, and diabetic retinopathy

Mayo Clinic

Manual abstraction vs. EMR-based algorithms: virtually all algorithms ultimately are dependent upon unstructured data; develop criteria for standardizing data dictionaries and best practices for handling missing data elements; community engagement survey instrument & educational video to educate community regarding biobank and community engagement processes; develop institutional policy and procedure for sharing of GWAS data; assess phenotyping heterogeneity from the EMR

Northwestern

Informatics: Identify shortcomings of data capture from routine clinical care and repurposed for research; Develop and implement common standards for formatting and sharing data; Community engagement: Develop model consent language; Summarize community engagement efforts around data sharing in our population; Genomics: Develop process for GWAS data certification review and approval; Other/general: Develop best practices for interacting with IRBs around biorepository formation and ongoing consultation

Group Health

Mapping the electronic derived cases vs. 'research quality' (e.g. dementia). How to handle cases from different sources; Use of "low tech" methods to extract NLP information; identify participant-centered best practices regarding consent from existing cohorts; develop recommendations for institutions, investigators re consent, data sharing, other issues with GWAS and related research (products from a consensus panel process)

Vanderbilt

Identify shortcomings and enhance internal EMRs to capture data in a structured format; develop methods for assessing/labeling certainty of data shared to public databases; create a description of the various analogs to human subjects biobanking in a non-human subjects model

Administrative Coordinating Center

Creation of a library of searchable phenotype algorithms plus associated metadata; creation of educational materials on genomic data privacy for IRBs and other regulatory decision makers; develop a re-identification risk framework for biomedical data to be shared to dbGaP