By Cindy Atoji
October 7, 2008 | McKesson is jockeying for a prominent position in the rapidly evolving personalized medicine race, developing capabilities for next-generation electronic health records (EHRs), says Andrew Mellin, vice president for Predictive Care Solutions. The San Francisco-based health care giant formed this new group last year within the Life Sciences division, focusing in particular on tweaking its Horizon Clinicals products—decision support and physician order entry—to support advancements in genomics, genetic testing, and bioinformatics as they emerge. “We view ourselves as a catalyst of personalized medicine,” says Mellin, who spoke to Digital HealthCare & Productivity about the company’s recent partnership with Proventys, and its strategies for personalized medicine.
DHP: What is McKesson’s strategy for personalized medicine?
MELLIN: We believe that you can apply many of the concepts of personalized medicine—having a more predictive and specific model for a given patient—even without a lot of genetic information. We’ve partnered with Proventys to help us develop risk models that will be incorporated into clinical decision making. We see this as an important incremental step even before genotyping becomes more pervasive; it’s a very similar concept.
We’re working very closely with Proventys to develop a series of predictive modules that will be incorporated into our clinical IT solutions. These include pharmacogenomic calculators and alerts that can help optimize and tailor a patient’s therapies. We’ll most likely start in the hospital setting with our physician order entry system, called Horizon Expert Orders, which today has already fairly sophisticated decision support tools that guide physicians toward complex decisions.
But what it doesn’t have today—and really no one has today—is [the ability to take] the patient info that already exists in the system, and automatically calculate a risk score that can help guide the physician to the most appropriate choice of therapeutic or diagnostic treatment.
We see concepts of personalized medicine expanding to many areas of the hospital, including areas of anesthesia complications, outpatient, predictive medicine, and beyond. The area that you see most in personalized medicine is oncology, but we also see opportunities in many common conditions, in cardiovascular disease, psychiatry, neurology, and even general internal medicine.
DHP: Is McKesson developing any other new capabilities for personalized clinical decision making?
MELLIN: The foundation of personalized medicine starts in the lab. The hospital lab systems are actually starting to do these genetic tests, and the pathologists are a key component in helping to understand the complex data and helping guide physicians to know what to do with this information. So we are enabling our core lab system to be able to manage, manipulate, and provide decision support for these genetic tests as they become more and more prevalent. We also will be enabling our core electronic health record to manage the genetic information that’s captured and maintained on a patient. It’s a tremendous amount of data that we need to be able to store and manipulate, visualize, and use for decision support in the electronic health record, so we are also exploring how to do that.
It’s an extremely complex topic. You’re not dealing with hemoglobin that’s a point-in-time single lab value, but a set of information that’s very large whose meaning changes over time as science advances. The EHR has to be able to think about and support this evolution of knowledge around genetic information.
And clearly there are other factors you need to think about, including privacy issues, and even things as simple as what you call the gene, and the way you store the gene. There is a lot of work we in the industry have to do to be able to fully embrace the concepts of personalized medicine.
DHP: How is McKesson working to incorporate genetic data standards as they emerge?
MELLIN: Our core clinical information systems have already been built in such a way that they incorporate knowledge models and knowledge management and are able to understand relationships between problems, key vocabulary terms, and other clinical findings. So it’s a natural extension take to be able to take a knowledge engine and incorporate these genetic standards as they emerge. Most genetic standards are focused on science and the research side of things so it’s not completely seamless to just take those standards and put them in the context of a clinical information system. There is work being done by standards organizations, which we are following closely, and as those standards finalize and emerge, that is how we will represent and store the data in our systems.
DHP: What about creating knowledge visualization and representation designs that support the use of genomic medicine?
MELLIN: We haven’t solved that problem yet. We’re thinking about it and looking at some concepts on how this data could be represented and used so physicians can cognitively interpret the data to help guide them to the right treatment. But we’re not there yet, and the reality is much of the complexity of the individual genetic data is going to have to be hidden. Visualization is important but even more important is selective visualization and showing the physicians the information they that need to know at the point of decision making.