VERMONT OXFORD NETWORK DATA INGESTION AUTOMATION

What is the Vermont Oxford Network

The Vermont Oxford Network (VON) is a worldwide registry for Neonatal Intensive Care Units (NICUs) working together as an interdisciplinary community to change the landscape of neonatal care. J P Systems assists VON by improving the quality of care for very low birth weight new-born infants. We do this by standardizing their patient records database though a coordinated program of data quality improvement, education and research. This detailed clinical data is now entered manually based on a questionnaire which affects infants in resource-limited settings all around the world. The goal of our project is to enable automated and internationally standardized data entry so data can be ingested faster.

Once standardized, we will be able to design FHIR™ apps which partially automate the workflow by ingesting data directly from vendor EHR systems (e.g., Cerner and Epic). Our data and terminology standardization efforts provide the neonatal community of practice the ability to achieve data-driven quality improvement.

We analyzed the data collected from the questionnaires which are sent out to the 1,200 member neonatal units all over the world.

Each question was broken down in to its elements so as to be able to match it to international reference terminologies. Some questions resulted in answers with simple terminological concepts. We were able to map them to a single SNOMED concept. Other questions resulted in compound concepts and were able to be mapped to two or more SNOMED concepts. The goal here, is to be able to identify the where the same data fields reside in the member hospitals' EHR systems. In the future a FHIR app will be able to pull data out of the EHR systems instead of having the hospitals fill out the questionnaires manually.

Yet, the answers to other questions were too complex to be mapped to fields in the EHRs. Rather, the answers are composed of drawing conclusions after researching multiple fields and treatment records. For example, some of the complex questions are; "Temperature within the first hour of admitting to the NICU," and "Bacterial Sepsis and/or Meningitis on or before Day 3." These questions are not mappable to individual EHR fields and require more study to automate the data collection process.