Can data analytics aid in end-of-life care decisions?

When it comes to clinicians discussing end-of-life care with patients and their families, too often decisions must be made quickly. The reasons why are complex: Physicians often struggle to determine when and how to have these tough conversations.

Nearly 70% of physicians report that they have not been trained to discuss end-of-life care, and 73% of Medicare patients over the age of 65 have not discussed it with physicians, according to a JAMA study released in November 2016.

HoganAs data analytics plays a larger role in healthcare, some are wondering how it might help address these difficult decisions. Yet most health systems don’t have the technology capabilities and the corporate mindset to take a comprehensive look at end-of-life care that better serves patients, says Dan Hogan, founder and CEO of Medalogix.

“Most hospitals aren’t using data analysts or data-driven tools to look at end-of-life care and imminent decline,” says Hogan, whose company specializes in population health analytics-based solutions for end-of-life care. He says that factors including increased readmissions, multi-episode hospital stays, and falls are part of a landscape of data that can help health systems identify patients who may need end-of-life conversations.

“Many times, once a patient gets to the emergency department they are already imminent,” Hogan says. “At that point, the doctors may try to stabilize them and get them out the door, or they may have already missed their mark.”

Hogan adds that the data also allows systems to look at population statistics and make predictions on which patients in the future need hospice care.

“There’s the benefit of averted costs to hospitals,” Hogan says. “The real benefit lies in patients and families being satisfied with how they and their loved one is treated. They are very grateful for a higher level of care during that time.”

Next: Untapped potential

Untapped potential

Data that could be used to create more accurate end-of-life timelines for patients is often unavailable to clinicians, says Ziad Obermeyer, MD, MPhil, assistant professor of emergency medicine at Brigham & Women's Hospital, assistant professor of healthcare policy at Harvard Medical School, and faculty affiliate of the Harvard Institute for Quantitative Social Science and Ariadne Labs at the Harvard School of Public Health.

“If you think about all the data living in an electronic medical record today, it’s an incredibly dense collection of lab tests, X-rays and CT scans, notes-a record of every point of contact you have with the medical system,” Obermeyer says. “It’s overwhelming, and hard for doctors to process. I think there’s a major role for algorithms here to digest and synthesize these complex data into usable information for both doctors and patients.”

In 2012, Obermeyer received the National Institutes of Health Director's Early Independence award to research patients who died unexpectedly after medical encounters. He realized that in many cases, too many factors were not being considering that led up to the patient’s seemingly quick death. For example, in many cases, patients who were candidates for palliative care were not receiving appropriate treatment.

“As I was researching these cases, I also found a lot of people who died after seeing doctors, but where death couldn’t have been unexpected: People with end-stage cancer, dementia, and other longstanding serious illnesses. And yet, these people were still getting very aggressive care, in and out of the hospital and emergency department,” Obermeyer says.

Obermeyer hopes more systems will be developed that can cross reference data to help empower clinicians to have these discussions with patients and their families earlier, when appropriate.

“Surveys of these patients show that they want to plan their legacy, get their affairs in order, and make decisions about how to spend their time,” Obermeyer says. “Doctors often don’t give them that information-because it’s really hard to have these conversations, but also because it’s really hard to predict. We’re going to see a big change in the availability of that kind of predictive information in the next few years.”

Next: Technology gaps

Technology gaps

Unfortunately, most electronic health record systems (EHRs) lack the ability to provide in-depth analysis of end-of-life factors, says Lee Goldberg, project director of improving end-of-life care for The Pew Charitable Trusts.

Goldberg“There’s a lot of variation and not a lot of standardization for end-of-life care. Some hospitals have customized modules in the EHRs. This can be very different based on location and very expensive to implement across an entire system,” Goldberg says, adding that poor care coordination and interoperability issues between EHRs can cause important documents to be overlooked. “Hospitals are important, but not the only players. The post-acute sector was not eligible for meaningful use dollars, and they typically lack the very robust EHR systems.”

Goldberg says that though most EHRs will prominently feature life or death illnesses or allergies, it can take up to 12 screens and 1.3 minutes to find advance care plans. The Care Planning Act sponsored by Sen. Johnny Isakson (R-Ga.) and Sen. Mark Warner (D-Va.) in 2015, would mandate that Medicare develop and test quality measures around end-of-life care, including care coordination between hospitals and other care settings.

Goldberg says that patients have more than three facility transitions in the last three months of their lives, and their advance care plans are sometimes misplaced.

“With some EHR systems, advance care plans can be uploaded in patient-generated fields. But a lot of other things may be in those fields, including Fitbit information, glucose monitoring, or data from another type of wearable device. Advance care plans can get lost,” Goldberg says.

Overlooked data source

There is another unused source of data that can help health systems improve the delivery of end-of-life care: surveys from bereaved families. Currently, Consumer Assessment of Healthcare Providers and Systems (CAHPS) surveys are for living patients. But only U.S. Veterans Health Administration hospitals give them to bereaved families.

These surveys ask families about care quality, and whether care was in accordance with the patient’s desires, says Goldberg. “We are hoping that [CMS] will create a CAHPS survey more like the VA,” he says. “A big question surrounds whether care received is what the patient wanted, and how are those wishes honored. Currently, there’s not a good quality measure for that, but technology and data mining could answer that.”

Donna Marbury is a writer in Columbus, Ohio.