Access Health Record at Push of Button When Evidence Lacking

Marcia Frellick

June 04, 2015

A green-button function being tested in electronic health records lets physicians access basic information on treatments and outcomes for patients similar to the one they're treating, say researchers at Stanford University in California.

The data being aggregated by the team will inform physicians at the point of care in the event that evidence from randomized trials is not available, said Christopher Longhurst, MD, from Stanford Children's Health, who is leading the green-button team.

The idea is to use practice-based evidence generated from clinical processes.

"There are always going to be situations where it doesn't make sense to randomize because populations are so small or the time required to see the end point would be too long," Dr Longhurst explained at the Big Data in Biomedicine Conference held at Stanford University.

The idea for the green button was generated after a 13-year-old girl nearing kidney failure was helicoptered to Lucile Packard Children's Hospital Stanford. After Jennifer Frankovich, MD, and her team, which included Dr Longhurst, diagnosed lupus, they debated whether the patient should be treated with anticoagulant therapy. Because lupus is not common in children, no evidence from randomized trials was available.

"In the adult world, you anticoagulate everybody who's not actively bleeding," Dr Longhurst said. "But in the pediatric world, the risk–benefit ratio is different."

Dr Frankovich did something that had never been done before. She used electronic health records to identify a cohort of 98 similar patients who had been treated at Stanford over the previous 5 years.

Leveraging Data

"Using some data in the text notes, she was able to identify an odds ratio of 7- to 15-fold greater relative risk of clotting for children like this one. Based on that, we made the decision to anticoagulate," said Dr Longhurst. A report on the experience was published the New England Journal of Medicine (2011;365:1758-1759).

If my daughter were hospitalized, I'd want her to get the most data-driven decisions, rather than what we do today.

We will never know if that was the right decision, but the girl didn't clot and didn't have adverse effects from the anticoagulant, Dr Longhurst said.

"If my daughter were hospitalized, I'd want her to get the most data-driven decisions, rather than what we do today," he said.

For such a system to be standardized, the privacy rule in the Health Insurance Portability and Accountability Act (HIPAA) of 1996 might have to change, Dr Longhurst said.

The system would need optimal cohort selection, and "we think the clinician should be part of that," he explained. Confounder control is also important. "When we know what the biases are, we can use tools such as high-dimensional propensity scoring to help control for those."

Also, the information would need to be integrated with the evidence to create true precision care.

A report by Dr Longhurst and his colleagues on what is needed to standardize this option at the bedside was recently published (J Comp Eff Res. 2015:11;1-7).

In the meantime, Stanford is taking the first steps toward such a system.

"The hospital compliance department, to their great credit, recognized that it made sense to allow our faculty members to search not just patients for whom they've had direct care contact, but all patients at Stanford Children's Health," he reported.

In the 3 months it's been live, it has changed care, said Dr Longhurst.

For example, "we were able to create a cohort of patients with cystic fibrosis who were receiving an antibiotic they shouldn't have been, and we immediately contacted them," he said. "And we were able to identify mothers with type 1 diabetes whose babies are going to be at higher risk for congenital heart disease and set up new screening."

Similar Systems

The green-button system is similar to Informatics for Integrating Biology and the Bedside (i2b2), said John Halamka, MD, from the Beth Israel Deaconess Medical Center and Harvard Medical School in Boston. i2b2 was developed at the National Center for Biomedical Computing, which is funded by the National Institutes of Health, and based at the Partners HealthCare System in Boston.

Any hospital can download the i2b2 software, Dr Halamka reported.

In his blog, he described the experience of using i2b2 to inform decisions about his wife's care after she was diagnosed with stage IIIA HER2-negative, estrogen-receptor-positive, progesterone-receptor-positive breast cancer. She is of Korean descent and was 50 years of age at the time.

He gathered information from i2b2 on what treatments similar patients had been given, and "she's completely cured," Dr Halamka reported.

Of course, the information hadn't been tested in a clinical trial, and just because 10,000 other patients receive a treatment doesn't mean it is the right treatment, he acknowledged. Therein lies the downside of such systems.

"But I think most physicians would be very interested in seeing, across the country, the current standard of care and the current standard practice," he pointed out.

The information isn't coming from hospitals to a central repository, which would be saddled with privacy issues. Instead, with i2B2, a question — such as, "Of all the Asian women who have had stage IIIA breast cancer, what chemotherapy did they receive?" — goes into the system. It's not disclosing any one patient's experience," he said.

Roadblocks to the current widespread use of these systems, therefore, is less about privacy and more about technologic capability and payment-model incentives, he explained.

Dr Longhurst reports serving on the medical advisory board for Doximity, a health IT startup; receiving research support from Hewlett Packard; and owning stock and stock options in Doximity. Dr Halamka has disclosed no relevant financial relationships.

Big Data in Biomedicine Conference. Presented May 22, 2015.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.