Ten years out, though, Evans can envision using genome sequencing even for preventive medicine, such as flagging people who have clear dispositions to develop a given disease. For example, a patient with the genomic variant for Lynch syndrome, which radically increases the risk of colon cancer, could be scheduled for a yearly colonoscopy and routine removal of precancerous polyps.
DiPersio, who cares for Lukas Wartman at Washington University, is cautious-not about the science itself, but about the snake oil purveyors who inevitably spring up around such a major scientific advance. He recalls the craze about 20 years ago for testing tumor tissues to create individualized chemotherapy regimens. A cottage industry of charlatans with shaky science-or none-gave false hope to cancer sufferers. "These tests are going to be cheap enough that you'll start seeing that approach," he says. "There will be the same army of guys delivering information that's either idiotic or confusing." He expects a significant time lag between research discoveries and the ability to use them to help patients.
Marc Williams, M.D., director of the Genomic Medicine Institute at Geisinger Health System, agrees that there's such a thing as too much, too soon. He points to a case where a patient used a direct-to-consumer genomic test to discover that he had an increased risk for prostate cancer, and then circulated among physicians until he found one willing to do a biopsy that showed a cancerous prostate. "He went around saying, 'This test saved my life!'" Williams says. "But if everyone did that, a lot of people would be treated for indolent cancers that would never have become a threat, and they'd be incontinent and impotent for nothing. We have to understand how to use this information effectively."
In many ways, genomic information is no different from any other personal health information, like a lab result or a family history. Its privacy needs to be protected, it has to be translated into a useful form for clinicians, it has to be stored, and it has to be shared.
The difference is primarily one of degree. An unusually large pile of data contains a few useful and potentially life-saving nuggets of information. And unlike most health information, the same pile of data can be mined for additional nuggets for at least the next several decades, as the genomic knowledge base grows.
Most providers will farm out genomic sequencing to independent labs or to large academic medical centers that may also handle the initial translation of the results into actionable information.
Baylor College of Medicine in Houston does exactly that for genomic tests, says Jeffrey Reid, assistant professor in its Human Genome Sequencing Center. The center reports what he calls "very digested data objects" that address only the issues that prompted the physician to order the sequencing. "Genetics does not yet do well with 'healthy people' testing, where there's nothing wrong with you or your genome, but there are signals that you have higher risk for breast cancer," he says. "Currently I think you can do that better with a very detailed family history. But that type of data will get better and better as more databases are built."
Health information technologists should be planning for a huge change within the next three to seven years, says Sandy Aronson, executive director of I.T. at the Partners HealthCare Center for Personalized Genetic Medicine. "This switch is going to happen very quickly when it happens, and it's important to start the process of planning how additional genetic support is going to be built into your electronic health record," he says. "Start planning now." Aronson and his team developed software called GeneInsight to provide exactly that type of support and have integrated it into EHRs used at both Partners and its research collaborator Intermountain Health Care.
Here are some things to look at in planning for EHRs for the genomics age:
Better family histories. "Family history has not been well represented in many EHRs," says Greg Feero, M.D., special advisor to the director for genomic medicine at the National Institutes of Health. He also practices family medicine in Maine, and his job is to help figure out how genomics will fit into everyday practice.
He says a more robust, structured family history could end up as a Stage 2 meaningful use requirement, specifically so that EHRs will be genome-ready.
Intermountain Health Care's Clinical Genetics Institute is developing an elaborate, Web-based family history tool that patients can use to build elaborate pedigrees, with each ancestors' diseases, date of onset, and cause of death. Grant Wood, the institute's senior information technology strategist, says all that information will go into the EHR, to help clinicians choose appropriate genetic and genomic tests and make sense of the results.
Enhanced clinical decision support. David Dimmock, M.D., the pediatric geneticist who identified the gene variant that plagued little Nicholas Volker, believes very little genomic information, as such, will end up in the EHR. Instead, he wants a built-in risk calculator that signals the physician when a pattern of family history and physical symptoms indicates that a genomic test might provide useful answers. Once a patient's genome is on file, its information should be available as needed.
"Our providers say they don't want a lot of data," he says. "They want just-in-time reminders. If they put in a diagnosis of depression, they want a flag that this or that drug would be better, given the person's genomic information."
Storage. Each genome takes up the equivalent of a terabyte drive. Though the medically relevant variations may eventually boil down to a much smaller file, it's still a lot of data. One strategy would be to store genomes the way a PACS stores images. But Intermountain's Grant Wood hopes to keep genome data from becoming one more silo.