Advancing care planning with a machine learning algorithm: One organization’s success story

BJC HealthCare focuses on targeted interventions along the care continuum to improve advanced care planning outcomes.

How can health organizations approach advanced care planning and make a positive impact? In this article, originally published by our sister publication, Healthcare Innovation, Contributing Senior Editor David Raths provides an overview of BJC HealthCare’s strategy.

Read a snippet of Healthcare Innovation’s article below.

How BJC HealthCare got better at advanced care planning discussions

“We’re in this odd situation in the U.S., where everybody agrees advanced care planning is super important, and nobody does it.” 

At the recent NAACOS meeting, that was how Nathan Moore, M.D., medical director of BJC HealthCare’s accountable care organization, led off his presentation about how BJC has increased advanced care planning and the impact it has had on the use of palliative care. 

St. Louis-based BJC HealthCare is a large integrated health system with 14 hospitals. Moore further explained the conundrum about the lack of progress on advanced care planning. He said reading the literature, two themes repeatedly emerge. One is that it's hard to figure out which patients need this intervention and when. Second, how do you engage providers to have these difficult, time-consuming conversations? “Primary care says you should talk to your oncologist. Oncologist says you should talk to your cardiologist,” he said, “Patient gets admitted, and the hospital says, ‘I just met you, you should talk to your primary care doctor.’ Nobody wants to be the bad guy.”

Some health systems’ approach is that advanced care planning should be like blood pressure — just do it for every patient, every visit. “That is not realistic, nor has it been shown to be helpful,” Moore said. “Instead, you have to figure out where along the care continuum you want to do the intervention. Is it outpatient, primary care, specialty care? Is it inpatient? Is it ICU? Is it post-discharge? Then you have to figure out which tranche of risk patients you are trying to target — who to target and when, and then who is providing the intervention.”

Providers and patients are both looking for excuses not to have this conversation, Moore stressed. “In our system, our mantra is make this as painless as possible.” He recommends focusing efforts on implementation and usability.

It also depends on what resources you have available in your health system. “But the real impact when it comes to ACOs and cost of care doesn't come from the conversation itself. It helps, but it's what you do after that,” Moore said. There's palliative care in an outpatient setting or telemedicine, and home-based palliative care.

He mentioned a publication about an ACO’s home-based palliative care reducing total cost of care substantially. In that case the palliative care was helpful, but what really moved the needle with their spending was that when those patients called their general triage line with dyspnea or nausea or fatigue, they didn't go down the typical protocols that everybody else in their triage line did. If they were already enrolled in home-based palliative care, they went down an alternative triage pathway so they weren't just reflexively sent to the emergency room. “You really have to change your downstream workflows with these patients to reduce utilization that we know isn't really going to help the patient,” Moore said. 

Another option involves hospital-at-home programs. Moore said he spoke with leaders of an ACO in Iowa a few years ago. They would identify their frail elderly that were likely to need admission in the next 12 months, and get everything set up for hospital at home for when the time came. Part of that process was thinking through how hospice is a likely outcome for a lot of these patients, so instead of being reactive, they were proactive and had those resources available. 

Then there is hospice. “I think all of us would agree hospice is the preferred pathway before dying for most patients, and the more you get the better — at least two weeks, and ideally closer to three to six months,” Moore said. 

Once you've decided on a program, what can you measure for success? You can track documentation in the EHR, hospice utilization and billing codes. 
“Our vendor automatically reports for our patients who die — did they get no hospice, hospice for less than two weeks, or hospice for more than two weeks,” Moore said. “There’s a substantial difference in total cost of care near the end of life. That's a fairly easy data point for ACO board members and hospital leadership to understand. I am happy to report that our ACO, for the first time ever, has finally hit national average for hospice use.”

Visit Healthcare Innovation for the full article.

About the Author

David Raths

David Raths is a Contributing Senior Editor for MLO sister brand Healthcare Innovation, focusing on clinical informatics, learning health systems and value-based care transformation. He has been interviewing health system CIOs and CMIOs since 2006.

Follow him on Twitter @DavidRaths

Sign up for our eNewsletters
Get the latest news and updates