The five-stage process for completing a legally valid living will & healthcare proxy — from readiness assessment to emergency-ready distribution
Advance care planning (ACP) works best when it begins long before a health crisis forces the conversation. Yet in US practice, advance directives are still disproportionately introduced in the emergency department or ICU — precisely the setting where rushed decisions, impaired capacity, and family distress make thoughtful planning hardest. The first stage of a well-run workflow is not paperwork at all: it is identifying whether this patient can meaningfully participate in the decision, and finding a calmer moment to start.
Advance directive completion in the United States has hovered around 33-37% of adults for over two decades despite decades of public health campaigns, National Healthcare Decisions Day, and electronic health record prompts. Meta-analyses consistently find that the single biggest predictor of completion is not education level or health status — it is whether a clinician ever raised the topic in a low-pressure setting.
The Medicare Annual Wellness Visit created two dedicated billing codes (CPT 99497 for the first 16-30 minutes of advance care planning discussion, 99498 for each additional 30 minutes) specifically to make room for this conversation outside of acute care. Other recommended "teachable moments" include: a new chronic disease diagnosis, a milestone birthday (65, when Medicare eligibility begins), transition into a nursing facility or hospice-eligible status, or after witnessing a family member's difficult end-of-life course.
Crisis-driven advance care planning — a resident asking a newly-admitted, acutely ill patient about code status within an hour of arrival — produces lower-quality decisions. Patients under acute stress, sedated, or in pain are worse at abstract, values-based reasoning, and the encounter often collapses into a binary "do you want everything done?" question that captures none of the nuance a living will is meant to convey.
Before any directive can be legally valid, the patient must have decisional capacity specific to this decision at this time. Capacity is not a global, permanent label ("demented" or "competent") — it is decision-specific and can fluctuate with delirium, sedation, pain, or the time of day. The standard clinical framework (Appelbaum, based on the MacCAT-T instrument) evaluates four domains:
1. Understanding — can the patient restate, in their own words, the relevant medical facts and the nature of the decision being made? 2. Appreciation — can the patient apply that information to their own situation (not just recite facts, but recognize "this applies to me")? 3. Reasoning — can the patient manipulate the information logically, weighing risks, benefits, and alternatives in a way that is internally consistent? 4. Expressing a choice — can the patient communicate a stable choice, even if that choice is unconventional?
A patient with mild-to-moderate dementia, aphasia from stroke, or fluctuating delirium may still retain capacity for this specific decision even if they lack capacity for, say, managing finances. Conversely, a cognitively intact patient in acute crisis (severe pain, panic, acute psychosis) may temporarily lack capacity and the conversation should be deferred or supplemented with a previously-designated surrogate.
When capacity is genuinely and permanently absent, the workflow shifts: the healthcare proxy or default surrogate (determined by state hierarchy statutes — typically spouse, then adult children, then parents, then siblings) becomes the primary decision-maker, guided by substituted judgment (what would this patient have wanted) rather than best-interest reasoning alone.
Multiple, overlapping barriers explain why advance directive completion has plateaued despite broad awareness:
• Clinician time and training — a meaningful ACP conversation takes 15-30 minutes; many primary care visits are scheduled for 15 minutes total, and many clinicians report inadequate communication training for discussing death and prognosis. • Discomfort and avoidance — both patients and clinicians often avoid the topic, patients out of superstition ("talking about it will make it happen") or fear, clinicians out of concern about damaging rapport or delivering bad news. • Health literacy — legal and medical terminology in standard forms ("artificially administered nutrition," "irreversible condition") is frequently above the reading level of the general population. • Distrust rooted in historical inequity — documented mistrust of the healthcare system among Black and Hispanic patients, shaped by events like the Tuskegee syphilis study and ongoing disparities in pain treatment and end-of-life care, contributes to markedly lower completion rates and a preference in some communities for family- or faith-based decision-making over written directives. • The "too healthy for this" assumption — both patients and clinicians frequently defer the conversation until serious illness appears, even though sudden incapacitating events (stroke, trauma, cardiac arrest) can strike healthy adults at any age.
National surveys consistently find advance directive completion near one-third of US adults, with rates roughly half as high among Black and Hispanic adults compared to white adults, and among adults with lower income and educational attainment — a disparity that tracks broader inequities in access to primary care where these conversations are most likely to happen.
A living will that lists treatments in isolation ("no CPR," "no ventilator") without grounding those choices in the patient's actual values is a blunt instrument. The second stage of the workflow uses validated, structured tools to walk patients through concrete hypothetical scenarios, helping them — and the people who will one day speak for them — understand not just what they want, but why.
Two tools dominate US values-clarification practice, each validated in different ways:
Five Wishes (Aging with Dignity, first published 1997) is a plain-language document covering five domains: (1) the person I want to make healthcare decisions for me, (2) the kind of medical treatment I want or don't want, (3) how comfortable I want to be, (4) how I want people to treat me, and (5) what I want my loved ones to know. It has been distributed to more than 40 million people and, because it was drafted with attorneys in each state, meets statutory advance directive requirements in roughly 42 states either as-is or with a state-specific addendum.
PREPARE for Your Care (Prepare for Your Care, developed by Dr. Rebecca Sudore at UCSF) is a freely available, multimedia, 5th-grade-reading-level program delivered online or on paper, explicitly designed for populations with limited health literacy. Unlike Five Wishes, PREPARE has been tested in multiple randomized controlled trials and shown not just to increase documentation rates but to improve patient-clinician communication about wishes and increase the likelihood that a chosen surrogate understands their role — an outcome that plain document completion alone does not guarantee.
Other tools in common use include the Go Wish card-sort exercise (patients rank cards describing values like "to be free of pain" or "to not be a burden to my family") and institution-specific worksheets, but Five Wishes and PREPARE remain the most widely validated and distributed nationally.
Rather than asking abstractly "do you want life support," structured tools present three canonical hypothetical health states, because patient preferences vary dramatically across them and a single blanket answer rarely fits all three:
• Permanent unconsciousness (persistent vegetative state) — no awareness, no chance of recovery, but the body can be kept alive indefinitely with artificial nutrition and ventilatory support if needed. Most patients, when asked specifically about this scenario, decline sustained life-prolonging treatment — but a meaningful minority want everything continued regardless, often for religious reasons. • Terminal illness with unrecoverable decline — a condition (e.g., metastatic cancer, end-stage organ failure) expected to cause death regardless of treatment, where life-sustaining interventions would only prolong the dying process. Preferences here often hinge on expected quality of remaining time versus duration. • Severe, irreversible dementia — the patient no longer recognizes family, cannot communicate, and requires total care, but is not imminently dying. This scenario produces the widest range of preferences and the most value-laden reasoning, since it is not "terminal" in the traditional sense — decisions often turn on whether continued biological life without recognizable selfhood is consistent with the person's sense of dignity and identity.
By walking through each scenario individually, values clarification tools reveal the underlying reasoning (independence, avoiding being a burden, cognitive connection to loved ones, spiritual beliefs about the sanctity of biological life) that a surrogate can then apply flexibly to situations the document never explicitly anticipated.
A substantial body of research, including foundational work by Sudore and Fried ("Redefining the Planning in Advance Care Planning," 2010), argues that advance care planning has historically over-invested in document completion and under-invested in the harder, more valuable work of values communication between patients and surrogates.
Treatment-specific checklists are inherently limited: they cannot anticipate every clinical scenario a patient will actually face, and rigid application of a checked box written years earlier — without understanding the values behind it — can lead a surrogate to make a choice the patient would not have wanted in the specific circumstances that actually arose. A form that says "no mechanical ventilation" completed by a healthy 50-year-old may not obviously apply to a three-day intubation with excellent prognosis for full recovery after a reversible pneumonia.
By contrast, when a surrogate understands the values underneath the checklist — "my mother cares most about being able to recognize her grandchildren and communicate, and is willing to accept significant physical limitation but not loss of cognitive connection" — they are equipped to reason through novel clinical situations the original document never mentioned. Research shows that structured values conversations between patients and their named surrogates measurably improve the surrogate's ability to predict the patient's actual preferences (surrogate-patient concordance), more so than legal-document completion alone.
Key Insight: The evidence base increasingly shows that the conversation is the intervention, and the document is merely its record. Programs that combine values clarification with structured surrogate conversations (like PREPARE) improve documented concordance between expressed wishes and eventual care more reliably than distributing a legal form in isolation.
Once values are clarified, they must be translated into a document with actual legal force. In the United States, this typically means combining two legal instruments — a living will, which states treatment preferences under specific medical conditions, and a healthcare power of attorney (healthcare proxy), which names a person authorized to make decisions the document does not anticipate. Roughly thirty states now provide a single combined advance directive form that merges both functions, though the underlying legal requirements still vary considerably by state.
A complete advance directive generally consists of two distinct instruments, which may be combined into one document or kept separate depending on state law:
1. Living will — a written statement of the patient's wishes regarding specific life-sustaining treatments under specific future medical conditions (typically "terminal condition" and/or "permanent unconsciousness," as defined by that state's statute). It takes effect only when the patient can no longer communicate and the triggering condition is certified, usually by one or two physicians.
2. Healthcare power of attorney / healthcare proxy (durable power of attorney for healthcare) — names a specific person (the "agent" or "proxy") with legal authority to make healthcare decisions on the patient's behalf whenever the patient lacks capacity, whether or not the specific situation was anticipated in the living will. This is generally considered the more flexible and important of the two instruments, because no document can anticipate every clinical scenario, and a well-briefed proxy can exercise substituted judgment where the living will is silent.
Because the proxy's authority is broader and more adaptable than a static checklist, most advance care planning experts now recommend that patients treat naming and thoroughly briefing a healthcare proxy as the single most important step in the entire process — even more important than the specific treatment boxes checked.
Standard living will forms ask patients to specify preferences for a defined set of life-sustaining interventions, usually under the triggering condition(s) defined by state law:
• Cardiopulmonary resuscitation (CPR) — chest compressions, defibrillation, and emergency medications after cardiac or respiratory arrest. This overlaps with, but is legally distinct from, a Do-Not-Resuscitate (DNR) order or Physician Orders for Life-Sustaining Treatment (POLST), which are actionable medical orders rather than statements of wishes. • Mechanical ventilation — breathing support via endotracheal tube or tracheostomy. Patients are typically asked to distinguish between short-term, potentially reversible ventilation (e.g., after surgery or treatable pneumonia) and indefinite ventilation with no reasonable expectation of recovery. • Artificially administered nutrition and hydration — feeding tubes (nasogastric or percutaneous/PEG) and IV fluids. This is consistently the most emotionally and legally fraught provision: it is often perceived less as "medical treatment" and more as basic care, and several state statutes require this preference to be stated with heightened specificity, sometimes in a separate signature or initialing step, precisely because courts have historically treated withdrawal of nutrition/hydration as a distinct legal question. • Dialysis, antibiotics, and other treatments — many combined forms also address renal dialysis, blood transfusion, and use of antibiotics for infections that arise during an otherwise terminal course, along with organ donation preferences and palliative/comfort-focused care instructions.
Because these provisions are often selected years before they might apply, most guidance recommends pairing each checked box with a short explanatory note in the patient's own words, so a future proxy or clinician understands the reasoning, not just the checkbox.
While every US state and DC has some statutory mechanism for advance directives, the specific legal requirements to make a document valid vary considerably:
• Witnessing requirements — most states require one or two adult witnesses to the signing; many exclude the patient's treating physician, employees of the care facility, and blood relatives or anyone who could financially benefit from the patient's estate, in order to prevent coercion or conflict of interest. • Notarization — a smaller number of states require notarization instead of, or in addition to, witnesses; the rise of remote online notarization (RON) since the COVID-19 pandemic has made this requirement considerably easier to satisfy in many jurisdictions. • Cross-state recognition — most states have "reciprocity" provisions that recognize an out-of-state directive validly executed under another state's law, but this is not universal or guaranteed, so patients who split time between states or relocate are advised to review their directive against the new state's requirements. • POLST/MOLST as a complement, not a substitute — Physician (or Medical) Orders for Life-Sustaining Treatment is a separate, brightly-colored, portable medical order form intended for patients with serious illness or frailty who are likely to face these decisions soon; unlike an advance directive, it is signed by a clinician and is immediately actionable by EMS and hospital staff, but it does not replace the broader values-based living will and proxy designation for patients who are not yet that sick.
A carefully drafted advance directive has no legal force until it is properly executed according to the formalities its state requires. This stage is where good intentions most often fail on a technicality: an ineligible witness, a missing notarization, or a signature obtained while the patient's capacity was already in question can render an otherwise thoughtful document unenforceable exactly when it is needed most.
Regardless of the specific state statute, valid execution of an advance directive generally requires: the document be signed (or, if the patient is physically unable to sign, directed to be signed by another person in the patient's presence and at their direction) by an adult who has decisional capacity at the moment of signing, the signing be a voluntary act free of coercion, and the document be dated.
Because capacity can fluctuate, many institutions recommend documenting a brief capacity assessment note contemporaneous with the signing — particularly for patients with early cognitive impairment, or those signing during a hospital admission where acute illness could raise later questions about the validity of the signature. A directive signed while capacity is genuinely in doubt is vulnerable to later challenge by family members who disagree with its contents.
Most states require the signing to be witnessed by one or two adults, and — critically — restrict who is allowed to serve as a witness. Common statutory exclusions include:
• The patient's treating physician or other direct healthcare provider — to avoid even the appearance that a clinician influenced a decision about the treatment they themselves would administer. • Employees of the healthcare facility where the patient is currently receiving care — for the same conflict-of-interest reasoning, particularly relevant for directives signed during a hospital or nursing-home stay. • The person named as the healthcare agent/proxy in the document itself — since that person stands to gain decision-making authority from the document, they cannot also attest to its valid execution. • Blood relatives, spouse, or anyone entitled to inherit from the patient's estate — to guard against financial conflicts of interest that could motivate pressuring a patient toward a particular choice.
These restrictions matter practically: a directive signed in a hospital room with two nurses as the only available witnesses may be invalid in a state that excludes facility employees, even though everyone involved acted in good faith. Patients executing a directive during a hospital stay should specifically request non-employee witnesses (visiting friends or family members not named as agent, or facility-designated patient advocates where available).
A minority of states require the directive to be notarized, either as an alternative to witnessing or in addition to it; some states offer patients a choice between the two paths. The expansion of remote online notarization (RON) statutes since the COVID-19 pandemic has significantly eased this requirement, allowing patients — including those who are homebound or hospitalized with limited visitor access — to complete notarization via video conference with an approved notary in many states.
Once validly executed, an advance directive generally remains in effect indefinitely until it is revoked or replaced — most states do not impose a hard expiration date, though a small number require periodic re-affirmation. Revocation is typically straightforward: the patient can revoke a directive at any time they have capacity, by destroying the document, executing a new one, or making a clear oral or written statement of intent to revoke, regardless of the formalities that were required to create it in the first place. This asymmetry — easy to revoke, more demanding to execute — is intentional, reflecting the legal principle that a patient's ability to change their mind should never be harder than their ability to state a preference in the first place.
The single most common reason a validly executed advance directive fails to guide actual care is not a legal defect — it is that nobody in the room during the crisis knows it exists or can locate a copy. The final stage of the workflow is arguably as important as drafting and execution combined: getting copies into every system and every hand that might need them, and revisiting the document as life circumstances change.
A signed advance directive sitting in a home filing cabinet or a single hospital's chart is only useful if the right people can retrieve it at the right moment — often during an emergency, at 2 a.m., at a hospital the patient has never visited before. A thorough distribution step routes copies to:
• The primary care provider — scanned into the outpatient chart, ideally flagged prominently so it is visible at the top of the record rather than buried in a documents tab. • The hospital or health system EHR — uploaded at the next admission or via patient portal upload where available; many EHRs support a discrete "advance directive on file" flag, but this flag is only as good as whether staff actually check it. • The named healthcare agent and close family members — who should have a physical or digital copy, not just awareness that one exists, since they may need to produce it quickly or explain its contents to an unfamiliar care team. • The patient themselves — a wallet card or phone-accessible copy noting the existence of the directive and the agent's contact information, useful if the crisis occurs away from any of the patient's usual care settings.
A persistent, well-documented problem in US healthcare is EHR fragmentation: a directive completed and uploaded at one health system is frequently invisible to a different hospital system the patient is later treated at, because most EHR platforms do not share advance directive data across institutional boundaries, even within the same metropolitan area.
To address the retrieval problem, roughly fifteen to twenty US states operate an official electronic advance directive registry, where patients or their attorneys can submit a copy of the executed document to a state-run database that hospitals and EMS can query in an emergency. Private national registries, such as the US Living Will Registry and other similar services, offer a parallel option for patients in states without an official registry, or as a supplement.
For patients who are already seriously ill or frail — not merely planning ahead, but likely to face an emergency decision soon — POLST registries (in states that operate them) serve a similar purpose specifically for the actionable medical orders that complement, rather than replace, the advance directive.
Despite these systems existing, actual registration rates remain low relative to completion rates, meaning even patients who have gone through the full readiness, values-clarification, drafting, and execution process often have not taken this final, low-effort step — leaving a validly executed document effectively invisible during exactly the kind of emergency it was written for.
Values and preferences are not static, and a directive written a decade ago by a healthy adult may not reflect that same person's preferences after a major diagnosis, a significant decline in function, or simply the passage of time and life experience. Advance care planning guidance commonly recommends review at the "5 D's": a new Decade of life, the Death of a loved one, Divorce or major change in family structure, a new Diagnosis of serious illness, and a significant Decline in health or functional status.
This matters because of a well-documented psychological phenomenon sometimes called the "disability paradox" or affective forecasting error: healthy people systematically underestimate the quality of life they would find acceptable once they actually experience a health state they currently consider undesirable. A patient who firmly rejects a scenario in the abstract may, upon actually reaching a milder version of it, value continued life more than their earlier document reflects — which is exactly why an engaged, well-briefed healthcare proxy who understands underlying values (not just checked boxes) is so important as a complement to the written document.
Research on documented advance directives generally shows that when a directive is both completed and actually accessible to the care team, it measurably improves concordance between a patient's expressed wishes and the care they ultimately receive, and reduces unwanted aggressive interventions at the end of life. But that benefit is conditional: a legally perfect document that cannot be found when needed provides none of that protection, which is why distribution and accessibility — often the most neglected stage of the workflow — deserves the same rigor as drafting and execution.
Key Insight: Studies of documented advance directives consistently find that patients whose directives are both completed and reliably accessible to their care team receive care more concordant with their expressed wishes and experience fewer unwanted invasive interventions at the end of life. The clinical benefit of an advance directive is therefore not created at the moment of signing — it is only realized at the moment someone can find and read it.