Last week, a multi-center study published in JAMA demonstrated the value of genetic testing for breast cancer risk, a simple test that allows women the choice of preventive surgery. The study and the accompanying editorial urged wider screening of women for the variants of the BRCA1 and BRCA2 genes that can create as much as ten times the risk of contracting breast cancer. But while a test for these risk variants has existed since the mid-1990s, only a fraction of women (and men) carrying the high-risk mutations have been identified. Some have blamed Myriad Genetics, the owners of the controversial patent on the BRCA 1 and 2 genes, for this shortfall. But what if a major contributing factor is something far less legally complicated: a physician’s finite time?
The average assessment of a patient’s genetic risk can take anywhere from 70 to 170 minutes, said Kevin Hughes, surgeon at Massachusetts General Hospital, who spoke to the University of Chicago Breast Cancer SPORE program Tuesday afternoon. Interviewing a patient about their symptoms and family history, running various risk assessment calculations and decisions, and producing the necessary paperwork eats up a great deal of time and effort even before the test is run. To find all of the approximately 1 million carriers of BRCA1/2 mutations in the United States, this exhaustive process would take an estimated 11,000 person years, Hughes calculated.
But a possible time-saving answer is to contract out some of those information-collection, decision-making, and paperwork-completing duties to a proven hard worker: the computer. Many of the electronic medical records systems rolling out in hospitals across the United States are ill-equipped to deal with the family histories and other information critical for genetic risk assessment, so Hughes and partners at Mass Gen developed their own free program, called Hughes RiskApps. A sort of electronic questionnaire that patients fill out on a tablet PC by checking boxes with a stylus, the program quickly gathers all the relevant information about the appearance of cancer in a patient’s relatives and their own medical history – but then, the real cool stuff happens.
Rather than dumping all the collected information on an overworked physician or genetic counselor, the program does its own thinking, a process known as Clinical Decision Support. By generating a family pedigree and running the various risk calculations itself, the computer program can make a recommendation in seconds on whether the patient – and other family members with a high risk – should be tested for BRCA mutations. What’s more, it can automatically generate all the paperwork necessary to obtain those tests, and even customized informational handouts to give the patient.
“The idea with this is to let the patient do the work putting the data in, let the computer do the work of doing the analysis, and then present it to the doctor in a way that makes it very easy for them to go on and take care of that patient and what to do next with them,” Hughes said. “It’s trying to take the workload away from the doctor and the center and putting it into the computer.”
Of course, key details still need to be checked and confirmed along the way by discussions between doctor and patient – the future is not a drive-thru ATM cancer risk assessment. But by speeding up many of the most time-intensive steps with the computer, Hughes hopes that more women at risk for BRCA1/2 mutations can be tested.
Still, expanding the system beyond the walls of Mass Gen will depend upon how the electronic medical record battle plays out. With the emphasis on electronic records in this year’s health care reform bill, more than 150 vendors have entered the race to provide the necessary technology. But many of these EMR systems aim to do everything, Hughes said, neglecting specific functions like family histories and cancer risk assessment that smaller programs do better. His solution is to push for interoperability – the ability to share data between records systems from different providers – and modularity, so that smaller “niche” programs can be plugged into broader EMR systems.
As medical care grows more and more complicated in the age of personalized genomics, saving time and stress with computational assistance looks more like a necessity than a luxury. But as Hughes’ talk emphasized, the computer doctors of our future need to be designed to reduce complexity, not add to it.