For the primary care providers in this pilot, advance care planning wasn't a conviction problem — it was a capacity problem. No room in a 20-minute visit to do the conversation well. No reliable way to see which patients needed it most. And a real cost to raising it cold with a patient who came in for something else.
So the work stayed undone. Of 150 attributed high-risk ACO patients, 3 had an advance directive on file.
Rosemary took the process off their plate — finding the patients, making the calls, completing the conversation, and returning documentation ready to review and sign. Six months later, 119 patients had been reached, 85 had completed an advance care planning conversation, and 71 had completed an advance directive and/or DNR.
The pilot included 150 attributed patients.
Only 3 patients had an advance directive on file. The other 147 did not.
These were not hypothetical future patients or names pulled from a broad outreach list. They were high-risk patients already attributed to the primary care organization.
The opportunity was sitting inside its existing patient population.
Reaching a patient is not the same as completing the work. Rosemary moved patients through the full advance care planning process.
Of the 147 patients without an advance directive on file, 119 were reached. During those conversations, Rosemary identified 25 patients who already had a current advance directive and 94 who did not.
That distinction matters because outreach is an activity. A completed conversation and documented care preferences are outcomes.
Rosemary connected with 119 of the 147 patients who did not have an advance directive on file.
Of reached, eligible patients, 85 of 94 completed a full advance care planning discussion.
71 of the 85 patients who completed the conversation also completed an advance directive and/or DNR.
Advance directives on file grew from 3 to 71 over the six-month pilot.
Rosemary contacted patients before an upcoming appointment and documented each advance care planning conversation in a structured form. An appointment note was placed in the EMR and tied to the patient's upcoming visit. The clinician's entire role was to review and sign through a browser-based workflow.
Nothing new to manage: no separate inbox, no separate system, no list to work between visits.
And no more raising ACP cold in the middle of a visit. By the time the patient arrived, the conversation had already happened — providers spent their minutes confirming decisions and answering questions instead of starting from zero.
"Almost no change"
Practice administrator, pilot site
One patient told Rosemary she had been carrying questions since her husband's hospitalization years earlier — no one had ever asked what she would want for herself. By the end of the call, she had named her daughter as her healthcare proxy and put her wishes in writing.
The program's economic model assigns $14,500 in avoidable last-year-of-life spending addressed to each completed advance care planning discussion, consistent with Aledade's study published in AJMC — $1.23 million in modeled avoidable spending addressed during a six-month pilot across three providers.
To be clear, this is not $1.23 million in audited or realized savings. It is modeled value tied to the potential difference between crisis-driven care and care guided by a patient's documented preferences.
Based on the conversion demonstrated during the pilot, expanding across the organization projects thousands of completed advance care planning discussions and tens of millions of dollars in avoidable spending addressed each year.
That is a projection, not an observed result. But it is grounded in a process that already worked across a defined patient population.
The result that matters most isn't a percentage. It's 71 people whose care, when it counts, is now far more likely to match what they actually want — and 71 families who won't have to guess.
That is what advance care planning is for. And it's exactly what stays undone when the conversation depends on finding room in a crowded visit.
Goal-concordant care isn't a conviction problem. It's a workflow problem — and the workflow now exists.
The fastest way to evaluate Rosemary is to be the patient. Pick a window and Rosemary will call you the way she calls them.
Pick a window →