“Data are trapped,” said Dr. Ned Sharpless, director of the National Cancer Institute. “This is a huge problem. It is phenomenally important.”

Share story

Dr. Nikhil Wagle thought he had a brilliant idea to advance research and patient care.

Wagle, an oncologist at the Dana Farber Cancer Institute in Boston, and his colleagues would build a huge database that linked cancer patients’ medical records, treatments and outcomes with their genetic backgrounds and the genetics of their tumors.

The database also would include patients’ own experiences. How ill did they feel with the treatments? What was their quality of life? The database would find patterns that would tell doctors what treatment was best for each patient and what patients might expect.

The holdup, he thought, would be finding patients. Instead, the real impediment turned out to be gathering their medical records.

In the United States, there is no single format used by all providers, and hospitals have no incentive to make it easy to transfer records from one place to another. The medical-records mess is hobbling research and impeding attempts to improve patient care.

“Data are trapped,” said Dr. Ned Sharpless, director of the National Cancer Institute. “This is a huge problem. It is phenomenally important.”

The cancer institute has invested millions of dollars into determining the genetic sequences of patients’ tumors, and researchers have found thousands of genes that seem to drive tumor growth.

But until patients’ medical records are linked to the genetic data, life-or-death questions cannot be answered.

“What drug did they get? Did they respond? Did they survive? Were they cured?” Sharpless asked.

The federal government has mandated uniform standards for electronic health records. “At this point, they are not to a level that helps with the detailed clinical data that we need for the scientific questions we want to ask,” Wagle said.

A few private companies are trying to tackle the problem. Flatiron Health, just bought by Roche, has obtained about 2.2 million records of cancer patients from medical centers and made them available for research after stripping them of identifiable information. But Flatiron must employ 900 nurses and certified tumor registrars, people with master’s degrees in coding data, to put it all into a usable form.

“About 50 percent, if not more, of the critical details we need for research are trapped in unstructured documents,” said Dr. Amy Abernethy, the company’s chief medical officer.

Wagle is making data from medical records and patients’ experiences public as he gets them. After 2 ½ years, though, he is disappointed by how little there is to share.

The patient who inspired his project had a lethal form of thyroid cancer. She was expected to die in a few months. In desperation, doctors gave her a drug that by all accounts should not have helped.

To everyone’s surprise, her tumors shrank to almost nothing, and she survived. She was an “extraordinary responder.”

Why? It turned out that her tumor had an unusual mutation that made it vulnerable to the drug.

And that got Wagle thinking. What if researchers had a database that would allow them to find these lucky patients, examine their tumors and discover genetic mutations that predict which drugs will work?

And what about those who were not helped by standard treatments? Could they be identified and spared treatments that will not work?

What researchers needed was a huge database that collected clinical and genetic data, along with patients’ descriptions of their experiences. Those narratives are crucial, Wagle said, but they are absent from the commercial databases like Flatiron’s. Those comprise anonymous patient data, making it impossible to ask the patients themselves how they fared.

Wagle decided to build a database, starting with metastatic breast cancer, his specialty. There are about 155,000 metastatic breast-cancer patients in the United States. He would use social media, online forums and advocacy groups to reach out to patients for their records.

Simply getting the records delivered, in whatever format, has been a nightmare. Records usually arrive as faxes or via snail mail.

“Even though the patients are saying, ‘I have consented for you to obtain my medical records,’ there is no good way to get them,” Wagle said.

He hired half a dozen people to work full time on the project, and corralled doctors and other experts to help part time. It can take hours to go through a single medical record.

Mary McGillicuddy, who works full time on the project, explained the system. When patients enroll, they tell the investigators where they were treated, where they had biopsies, where they had scans, and where they had medical procedures.

They give McGillicuddy and her colleagues permission to request their records. McGillicuddy faxes requests for records to each medical institution that treated a patient, or diagnosed or sequenced her cancer.

Startlingly, faxing “is the standard,” she said, for medical-records requests.

The process can be frustrating. Fax numbers can be out of date. Some medical centers will not accept electronic patient signatures on the permission forms.

So far, the breast-cancer project has received 450 records for 375 patients. (Each patient tends to have more than one record, because the women are typically seen at more than one medical center.)

“Patients are incredibly engaged and excited,” Wagle said. But for the records problem, “right now there isn’t a good solution.”