County health officials recently asked Duke how many of its patients would need priority access to the H1N1 flu vaccine. Duke used IBM Cognos to sift through information on the more than 20 million patients in its Oracle-based clinical data repository and in an hour was able to identify about 120,000 of them with risk factors, such as age, pregnancy, respiratory, and other conditions that made them vulnerable to complications from Swine flu. And now that the H1N1 vaccine is available, Duke is letting those patients know that they're first in line to get it.
"We put an analytics engine on top of our clinical repository and were able to stratify by age and key illnesses millions of records, and streamline who was most at risk," says Asif Ahmed, diagnostics services CIO for the Duke system, which runs three hospitals and about 100 clinics in the Raleigh/Durham, N.C., area and treats more than 1 million patients a year.
This is a practical example of how healthcare IT is being used to personalize medical care in ways that help doctors make smarter decisions about patients' conditions and tailor treatment to an individual's needs. This evolving field covers a broad range of efforts. Beyond analytics systems like Duke's, it includes decision-support tools that help doctors pick the best tests and treatments for patients, remote monitoring tools that provide close to real-time care, as well as software that helps researchers identify the best candidates to participate in trials or experimental treatments.
The Right Diagnosis
At Beth Israel Deaconess Medical Center in Boston, helping doctors make better treatment choices and arrive at more accurate diagnoses is a big and growing area of personalized medicine. One example is clinical support software to help its 1,600 staff and affiliated physicians choose the best radiology tests for patients.
When ordering CT scans, MRIs, X-rays, ultrasounds, and other radiology tests, doctors enter a patient's electronic medical record number into the Anvita Health decision support system. Data from Beth Israel's records system, such as recent lab tests and allergies, is automatically loaded into the software. The doctor then adds information on the current complaint, such as symptoms, what area of the body is a concern, and the suspected diagnosis as well as whether the person has any implants that might interfere with radiology treatment.
The software analyzes the data and rates the best tests for the patient, giving five stars for the top choices and one for the worst ones based on the risks and benefits of each. It also can recommend that the patient forgo radiology testing.
The system can catch details that might otherwise elude a doctor, such as a previous blood test indicating decreased kidney function that could mean the patient can't metabolize the dyes used in certain radiological tests. It also checks how much radiation the patient has already been exposed to.
"Excessive radiation can cause second malignancies," says Dr. Richard Parker, medical director of Beth Israel's physician organization. "The system takes that into account when ordering a scan." For instance, the software might point out that a patient suspected of having pneumonia has enough symptoms and clinical indicators to make that the most likely diagnosis, and that treating the patient for it would be better than exposing him or her to a chest X-ray.
Beth Israel launched a related pilot project six months ago to analyze doctors' thought processes when ordering radiology tests. When a doctor orders a test, the system asks what diagnosis the physician is leaning toward, with what percentage of certainty. After the test, the system follows up with an e-mail asking the doctor whether the test confirmed the original diagnosis. The study aims at gaining insight into how doctors decide which tests to use, and learning in which situations doctors are most likely to prescribe the wrong test for a given set of symptoms.
Information technology isn't just helping doctors choose the right test for a patient, it's also making more personalized medical tests possible. For example, diagnostic testing services provider Quest Diagnostics and Vermillion, a molecular diagnostic test developer, have developed a test to assess the likelihood that women diagnosed with pelvic masses have ovarian cancer as opposed to benign tumors. Its use is helping get women most at risk for cancer to specialists faster (see story, "Quest Stops Killer With Specialized Diagnostic Software").
|The Vision||The Challenges|