The New York Times Points Out How Useless The myHR Will Be For Serious Clinical Research.

This appeared a few days ago.

New Cancer Treatments Lie Hidden Under Mountains of Paperwork

By Gina Kolata

May 21, 2018

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

Dr. 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 would also 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?” Dr. 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,” Dr. 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.

“They are in PDFs. Maybe a doctor put in a note by hand, maybe a doctor typed it. That note became a narrative. It is not something that can easily be put into a spreadsheet.”

Dr. Sharpless worries that the data acquired by companies like Flatiron will not be readily available to researchers. But if the companies manage to solve the medical records problem cheaply, he said, “we’d like to work with them to figure out how to liberate the data.”

Dr. Wagle is making data from medical records and patients’ experiences public as he gets them. After 2 1/2 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.”

Vastly more here:

https://www.nytimes.com/2018/05/21/health/medical-records-cancer.html

Much the same with the “data are trapped” will happen with the myHR

The myHR, as we all know, is a mountain of .pdfs with some atomic data and images etc. to handle pathology and imaging results. The closest to atomic data will be the medicines data I would suspect.

Data at the level described above may be partially captured but will more than likely only even be partially available in practice systems.

Sadly I suspect the solution of re-entering the important data is going to be the only viable one as it has turned out to be in the U.S. at a cost I suspect no one will be able to afford.

So, what sounds really simple turns out to be just too hard!

David.