Talking About What Matters in Serious Illness

What is a goals of care conversation?

A goals of care conversation is a structured discussion about three things: what the illness is doing and is likely to do, what treatment can and cannot achieve, and what the sick person values enough to organize care around. The output is not a form. It is a shared understanding that lets every later medical decision be tested against one question — does this serve what the person said mattered?

The conversation sits inside the larger practice of palliative care, which the World Health Organization defines as an approach that improves quality of life for patients and families facing life-threatening illness through the prevention and relief of suffering — one that affirms life, treats dying as a normal process, and intends neither to hasten nor to postpone death 1Ref 1World Health Organization (2020).Palliative care.The WHO definition of palliative care as an approach improving quality of life for patients and families facing life-threatening illness through prevention and relief of suffering, affirming life, regarding dying as a normal process, and intending neither to hasten nor postpone death.. That framing matters, because a goals of care conversation is not a death conversation. It is about living with serious illness on the person's own terms, for however long that is. The National Institute on Aging treats these decisions — comfort, place of care, who decides — as the core of preparing for the end of life, well before the end is close 2Ref 2National Institute on Aging (NIH) (2022).End of Life.That end-of-life preparation covers decisions about comfort, place of care, and decision-making, and that advance directives and health care proxies are part of that preparation, per NIA's consumer end-of-life materials..

Three things a goals of care conversation is not:

• A one-time event. Goals shift as the illness shifts. The conversation is redone at every major turn.
• A hospice referral. Naming what matters does not enroll anyone in anything.
• A verdict. Nothing said in it is binding forever; people change their minds, and the record changes with them.

When should the conversation happen?

Earlier than almost anyone schedules it. The natural moments are a new serious diagnosis, a major hospitalization, a treatment crossroads, or a visible step down in what the person can do in a day. Waiting has a specific cost: decisions that were never discussed get made anyway — in an emergency department, by a family member guessing.

Professional bodies have started writing this into standard care. After a severe stroke, for example, the American Heart Association and American Stroke Association's statement for clinicians builds goals-of-care discussion into the care of the patient precisely because the course after a major stroke is hard to predict — some people recover function for months, others decline quickly, and the honest answer at the bedside is often a range 3Ref 3American Heart Association / American Stroke Association (2014).Palliative and End-of-Life Care in Stroke: A Statement for Healthcare Professionals From the American Heart Association/American Stroke Association.That AHA/ASA guidance incorporates goals-of-care discussion into care after severe stroke and acknowledges prognostic uncertainty in the post-stroke course..

Timing confusion is also where hospice anxiety creeps in. A goals of care conversation frequently ends with more treatment, not less — a decision to pursue a therapy aggressively is a goal of care too. And for people who are seriously ill but not near death, the conversation often clarifies exactly where they stand. Families frequently leave relieved to have named their actual situation — including when it isn't hospice yet, and no one is suggesting otherwise.

How do clinicians structure the conversation?

Good clinicians do not wing this. Most use some structured approach, and the structures share a spine: ask before telling, surface values before options, and check understanding before deciding. The Agency for Healthcare Research and Quality's SHARE Approach names five steps — seek the patient's participation, help the patient explore and compare options, assess the patient's values and preferences, reach a decision together, and evaluate that decision — with the explicit instruction to explore benefits, harms, and what matters most to this particular patient 4Ref 4Agency for Healthcare Research and Quality (2020).The SHARE Approach.The five named steps of the SHARE Approach to shared decision making and its instruction to explore benefits, harms, and what matters most to the patient, including assessing values and preferences..

Many palliative teams also work from a serious illness conversation guide, a script of open questions that keeps the talk on values rather than procedures. Families can ask whether the clinician uses one; the question itself signals that a real conversation is wanted, not a consent form.

Plain language is not a courtesy here — it is a safety practice. AHRQ's health-literacy toolkit recommends "universal precautions": structure every conversation so that anyone can follow it, regardless of background, using plain words and the teach-back technique, where the clinician asks the patient to say the plan back in their own words to expose gaps 5Ref 5Agency for Healthcare Research and Quality (2024).Health Literacy Universal Precautions Toolkit, 3rd Edition.The universal-precautions approach to health communication — structuring conversations so all patients can understand them regardless of literacy — and the plain-language and teach-back techniques the AHRQ toolkit recommends.. Worth asking the clinician to slow down and use teach-back if the words are sliding past. That is a normal request, not a confession.

What questions surface what matters most?

The useful questions are concrete and slightly uncomfortable, and they trade medical vocabulary for the shape of a day. Clinicians ask versions of them because the answers convert an abstract wish — "I want to fight this" — into information a treatment plan can actually honor. Asked at the kitchen table before the appointment, they work just as well:

• What does a good day look like now? What would make more days like that?
• What are you hoping for — and what else, if that first hope isn't possible?
• What are you most afraid of? Pain, dependence, being a burden, dying away from home — the answers differ more than families expect.
• What would you be willing to go through for the chance of more time? What would you not?
• Is there anything you need to do, finish, or say — and does the timing of treatment protect it?
• Who speaks for you if you cannot speak for yourself, and do they know what you just said?

Some families spend time naming what matters most before the appointment, so the person's answers are theirs rather than improvised under fluorescent lights. Trade-offs get real at this level of detail: an aggressive treatment schedule reads differently when fatigue in serious illness already takes most of the day, and saying that out loud is exactly the kind of information the SHARE Approach's values step is designed to capture 4Ref 4Agency for Healthcare Research and Quality (2020).The SHARE Approach.The five named steps of the SHARE Approach to shared decision making and its instruction to explore benefits, harms, and what matters most to the patient, including assessing values and preferences..

One caution: these questions are for the sick person to answer, not for the family to answer about them. The most common failure mode of a goals of care conversation is that everyone in the room speaks except the patient.

How do you make sense of prognosis and the numbers?

Ask for absolute numbers, in plain frames, and expect honest uncertainty. Decades of risk-communication research show that people — including clinicians — understand "10 out of 100 people like this" far better than percentages and conditional probabilities, that absolute risk is more honest than relative risk, and that survival-rate framing can mislead in ways mortality framing does not 6Ref 6Gigerenzer G, Gaissmaier W, Kurz-Milcke E, Schwartz LM, Woloshin S (2007).Helping Doctors and Patients Make Sense of Health Statistics.That natural frequencies communicate risk better than conditional probabilities, that absolute risks are clearer than relative risks, and that survival-rate framing misleads where mortality framing does not — for clinicians and patients alike.. A "50 percent improvement" can mean two extra people out of a hundred. It is fair to ask which.

Three questions that convert statistics into something usable:

• Out of 100 people in my situation, what happens to them? Natural frequencies, not percentages 6Ref 6Gigerenzer G, Gaissmaier W, Kurz-Milcke E, Schwartz LM, Woloshin S (2007).Helping Doctors and Patients Make Sense of Health Statistics.That natural frequencies communicate risk better than conditional probabilities, that absolute risks are clearer than relative risks, and that survival-rate framing misleads where mortality framing does not — for clinicians and patients alike..
• Is that number about people like me — my age, my other conditions — or about everyone with this diagnosis?
• What does the range look like? Prognosis is an estimate with wide error bars, and clinical guidance in conditions like severe stroke openly builds that unpredictability into its recommendations 3Ref 3American Heart Association / American Stroke Association (2014).Palliative and End-of-Life Care in Stroke: A Statement for Healthcare Professionals From the American Heart Association/American Stroke Association.That AHA/ASA guidance incorporates goals-of-care discussion into care after severe stroke and acknowledges prognostic uncertainty in the post-stroke course..

Another frame many families find useful: asking the clinician to describe the best case, the worst case, and the most likely case. The three-scenario version carries the uncertainty honestly — it plans for a range instead of an average — and it gives the family shared language for the decisions that arrive later.

"I don't know" from a clinician is not evasion. It is often the most accurate sentence in the room, and a good conversation plans for more than one future rather than betting everything on the average.

What if the family disagrees, or the person won't talk?

Two different problems, two different moves. When the family disagrees, the anchor is the person's own voice: the question is never "what do we want for Dad" but "what did Dad say he wanted" — and if nobody knows, that is the gap to close, not a vote to hold. Much family conflict in these rooms is grief wearing the costume of a medical opinion. Naming that gently — "we're all scared; let's talk about what he told us" — moves more than debating.

When the person themselves will not engage, pushing rarely works. Some people delegate — "do whatever the doctor thinks" — and that is itself a stated preference worth writing down. Others open up sideways: about a friend's death, about what they watched a parent go through. Those comments are data. A clinician-led family meeting can also carry weight that a kitchen-table conversation cannot; asking the care team to convene one is a standard request, not an escalation.

What does not help: springing the conversation during a crisis, holding it about the person rather than with them, or treating a first refusal as final. People often need the subject raised more than once, by more than one voice, before they are ready to answer.

What happens after the conversation?

The conversation only protects anyone if it leaves the room. What was said needs to become a chart note, a set of documents, and a story the family can retell — because the people who act on it later, in a hospital at night, will mostly be people who never heard it. Three durable outputs:

• A record in the chart. A direct request works: "Will you document what we discussed today, and where?" Goals that live only in memory get re-litigated at 2 a.m. by staff who never met the person.
• Documents. This is where advance care planning begins in earnest — a health care proxy or durable power of attorney naming who decides, and an advance directive describing what the person does and does not want. The National Institute on Aging's end-of-life materials walk through these decisions and the settings where they play out 2Ref 2National Institute on Aging (NIH) (2022).End of Life.That end-of-life preparation covers decisions about comfort, place of care, and decision-making, and that advance directives and health care proxies are part of that preparation, per NIA's consumer end-of-life materials..
• A shared story. The proxy and close family need to hear the person's words, not a summary. A proxy who can quote — "she said she'd trade time for being home" — is far harder to argue with in a hallway.

Then it gets revisited. New diagnosis, new hospitalization, new decline, new year — any of these is a reason to ask whether the goals still hold. Most of the time the answer is yes, and the re-asking takes ten minutes. The times the answer is no are exactly the times the conversation earns its keep.