In a pathology informatics webinar delivered yesterday by Dr. Mike Becich and presented by API and Sunquest (see: Free Informatics Webinar Tomorrow: IT Support for Pathology Research), I posed the question whether he knew of any cases of a "reverse feed" of clinical information from an EHR to an LIS. Ul Balis was the first informaticist who had used the term "reverse feed" in my presence but I am not sure if he originated it. A feed of clinical information to the LIS will be absolutely necessary for pathology to fulfill its evolving mission in molecular diagnostics and cancer genomics because it enables pathologists to refine their diagnoses and generate therapeutic recommendations. Pathologists obviously have manual access to the EHR but the volume of such data demanded by molecular and genomic pathology requires an electronic interface back to the LIS.
Dr. Becich said that he knew of no instances of reverse feeds of clinical information from the EHR to the LIS in any hospital. In my opinion, there are a number of reasons why such a reverse feed will never be allowed by EHR vendors. Here are some of the reasons for my statement:
- The LIS, RIS, and PACS systems are viewed as "ancillary systems" with the sole purpose of sending diagnostic data to the EHR with which the clinicians interact. For an EHR vendor, there would be no rationale or justification for such a "reverse feed" to the LIS from a competitive or business point of view. They would never articulate such an argument, however. Instead they would say that there is no need for such an interface given the integrated nature of the EHR database. They would omit the fact that there are few available tools to perform "deep phenotyping" studies on the EHR. Below is a definition of deep phenotyping for precision medicine (see: Deep Phenotyping for Precision Medicine)
- Deep phenotyping can be defined as the precise and comprehensive analysis of phenotypic abnormalities in which the individual components of the phenotype are observed and described....The comprehensive discovery of such subclasses, as well as the translation of this knowledge into clinical care, will depend critically upon computational resources to capture, store, and exchange phenotypic data, and upon sophisticated algorithms to integrate it with genomic variation, omics profiles, and other clinical information.
- An EHR vendor like Epic offers an enterprise-wide-solution with its own LIS and RIS. The company would thus not be inclined to participate in any process that enhances the functionality of a best-of-breed LISs like Sunquest, Soft, or Cerner and, in so doing, enhance their perceived value.
- An EHR vendor will assume that any algorithms used for deep phenotyping would run on its own system. Although such processing would be very desirable, I don't personally think we will see the development of such algorithms in the foreseeable future for EHRs. Although EHRs store massive amounts of clinical data, they are destined to function primarily as archives of clinical data without advanced data processing features.
Dr. Becich went on to state that very sophisticated data integration and analysis (i.e., deep phenotyping) is now occurring within the Department of Pathology at Pitt. The clinical data that the department of pathology requires to support its clinical and research mission is obtained from medical center's clinical data warehouse or repository which is populated with data by the EHR (including lab data) and to which the LIS is interfaced. At the end of the day, this may well be the best architecture for hospitals and medical schools because it spares the EHR the cycle burden of supporting interfaces back to the ancillary systems like the LIS and RIS. The downside of such an architecture is that the creation of a clinical data warehouse will probably be limited to the larger academic medical centers that are able to justify the expense by its research mission and the funds generated from it. Smaller hospitals will thus be deprived of the opportunity to integrate pathology test results with the relevant clinical data.