Observational Health Data Sciences and Informatics

The general motivation for creation of the OHDSI network is to conduct retrospective clinical research on observational data in a systematic and reproducible manner and at a scale that is unachievable today. It seeks to scale comprehensively across entire medical domains (e.g. all drugs for their effects, all diseases for certain attributes, etc.) as well as across different types of data and healthcare settings (claims, EHR, international). It is not designed for prospective research requiring investigator identification, patient recruitment, randomization. It is also not a service organization, but instead seeks to conduct research for the purpose of gaining insights and knowledge.

The network consists of Data Nodes and the Research Center. All research is originated and executed by the Nodes, while the Research Center provides the enabling logical network infrastructure (websites, sandbox for development and testing of code, etc.).

Each research question is addressed within the scope of a “Study”. A Study consists of:

a. Generation of a research question, which is “marketed” throughout the network to gauge if the question has potential to generate valuable insights. 
b. Generation of the protocol and executable code.
c. Execution of the code at the Data Nodes (could be a single Node).
d. Aggregation of the results and delivery of the conclusions (publication or otherwise).

If participating in the Network as a Data Node, institutions would enjoy participation in clinical research that would otherwise be unreachable due to lack of expertise (clinical, technical), scale, or mandate.

• Publications with leading scientists in the field
• Research grants (public or private sector) otherwise unachievable
• Public visibility for the institution

In addition, the Data Node is entitled to initiate Studies for the Network.

OHDSI is inviting institutions to join as a participating Data Node under the following conditions:

• The institution has data assets for conducting retrospective clinical research.
• The institution has the organizational mechanisms to participate in such research. This means skilled personnel with the necessary mandate and time. It also may or may not require the ability to interact with an ERB or Ethics Committee.

To participate, the institution would agree to:

a. Convert the data to CDM and keep them up to date
b. Participate in the Studies. 

(a) OHDSI will review the result of (a) and (b) and issue a certificate if the necessary quality and data characterization artifacts are delivered.

(b) is the condition which might cause stress for some Data Nodes, either because of limited resources or because the nature of a Study may prevent the Data Node from participation for ethical, commercial or political reasons. If a Data Node opts out from running a study, this fact and the reason category will be published through the Research Center.

Institutions without data assets can participate as clinical researchers, statisticians and programmers. There are no conditions for such participation.