OHDSI Home | Forums | Wiki | Github

Advocacy for Quick Win - EHR Export-To OMOP CDM

Many experienced leaders in healthcare community, who are yet to fully capitalize the benefits of OHDSI through adoption of OMOP CDM, remember the days when Adobe introduced PDF and how this transformed document sharing or how challenging it was before the days of being unable to share data without a common data format such as CSV across databases.

In Page 21 of “Peer Review of a Report on Strategies to Improve Patient Safety”, National Academy of Sciences’ committee recognizes that [now] “more than 90 percent of physicians and more than 95 percent of hospitals routinely collect data in an EHR. This development fundamentally changes what is feasible and practical, and the committee believes it should heavily influence strategies to improve patient safety going forward.”

Given the very high adoption of EHR, recognizing the commitment, contributions and possibilities of OMOP to generate reliable evidence, I am envisioning the day when the EHR Vendors would have a plug-in/app (as a paid feature or a free feature) or an “Export-to” option for OMOP CDM (which would take of all pertinent elements such as Vocabulary, GITHUB DDLs) instead of making it an intentional (and sometimes arduous effort even with all the help from OHDSI Community) effort by the health care organizations. This will be a substantial quick win for OMOP CDM adoption.

I am sure we have some pioneer EHR Vendors provide this already or could be lobbied by OHDSI Executive Leadership and by OMOP Ambassadors (all of us) using an appeal to their EHR Vendors requesting for an end-to-end jump start enabling feature.

Meantime, I am new to OHDSI and reviewing the resources to equip myself, advocate, and implement OMOP CDM within my organization.

If a direct EHR Export-To OMOP CDM already exists or is in development (to some extent), I will appreciate reference documentation on this, specific to Cerner EHR or Epic EHR.

Thank you so much,

[ Please note that the ideas are not expressed on behalf of my employer and my personal wish to community at large]


National Academies of Sciences, Engineering, and Medicine 2021. Peer Review of a Report on Strategies to Improve Patient Safety . Washington, DC: The National Academies Press. [https://doi.org/10.17226/26136]

Welcome to the journey, @ragha123!

I look forward to that day, too! However, there are quite a few hurdles to get over before we get there. The largest obstacle is the customization of the EHR system during implementation. Most hospitals/healthcare systems allow their care areas or departments to chose how they want their data displayed for the UI or reporting purposes. Different displays for different healthcare processes requires different backend storage of the data. I just recently dove into the storage of Covid vaccine data in Caboodle, the abstracted, dimensional model of Epic Clarity. I found Covid vaccine data in 5 different Caboodle tables. Unfortunately, these data are not directly linkable between the tables via a PK/FK relationship. These data are also coded with different vocabularies depending on the table storing the data. NDC, RxNorm, CPT4, ICD10PCS, and free text string (my personal favorite!) are all used to encode the data. Per current CDC guidelines, a Person will receive a maximum of 2 Covid vaccine drug exposures, so when I query a Person’s Covid vaccine records and receive a total of 8 records (all different source codes mapping to different standard concept_ids), I know duplication of the source data are rampant. And this duplication throughout the backend of the EHR and coding with varying levels of granularity are caused by the differing implementation strategies requested by the EHR vendor’s customer. And what is true for Covid vaccines is true for all domains in the EHR. Meaningful Use gave incentives to implement the EHR, but did not provide instructions or guidance for the actual implementation of the EHR.

Because of the different ways an EHR can store data on the backend, there isn’t a one size fits all ETL from EHR to OMOP. However, there are some resources within and outside the OHDSI community.

  • Epic User Web has some ETL scripts and discussion. Due to the proprietary nature of Epic’s EHR, this is only accessible to Epic customers

  • Here’s the link to an older Cerner ETL

  • You found your way to the EHR WG this morning. We provide support for all EHR data by giving guidance through the ETL process, maintaining lists of current resources, encouraging collaboration, and possibly establishing a mentor program between veteran EHR OMOPers and those just joining the journey. The WG hasn’t established a mentoring program yet, but I will add it as one of the discussion items for an upcoming call. If we are able to establish a mentoring program, then it could possibly support the various EHR systems. Depending on the knowledge of the mentors.

The above are all starting points for an ETL conversion from EHR data. I highly suggest you also engage reporting experts, EHR experts, and those with access to the UI to assist in the transformation.