Assisted dying is an umbrella term for practices in which another person, usually a health professional, intentionally helps bring about an individual’s death at that individual’s voluntary request. Its meaning varies between jurisdictions and institutions. In some settings it refers only to the prescription of medication that the person administers themselves; elsewhere it also includes administration by a physician or another authorized clinician. These differences are substantive: they determine who performs the final act, which professionals may participate, and what legal safeguards apply.
The subject lies at the intersection of medicine, law, and ethics. It concerns the relief of suffering, individual decision-making, professional responsibilities, and the protection of people who may experience coercion or disadvantage. It is distinct from refusing medical treatment, withdrawing unwanted life support, and providing proportionate symptom relief near death, although all may arise within end-of-life care.
Assisted dying has no single international legal or clinical model. Laws differ over terminal illness, the nature of qualifying suffering, age, residency, decision-making capacity, and oversight. Understanding the subject therefore requires attention to the terminology used, the precise legal framework, the assessment process, and the limits of the available evidence.
Terminology and conceptual boundaries
The principal distinction is between self-administered and clinician-administered practices. In physician-assisted suicide, a physician supplies or prescribes medication with the understanding that an eligible person may use it to end their own life; the person performs the final act. Some statutes and professional organizations instead use terms such as medical aid in dying or physician-assisted dying. These labels are not interchangeable across legal systems, and the word “suicide” itself is contested in this context. Some jurisdictions expressly distinguish deaths under their assisted-dying statutes from suicide for specified legal purposes.
In euthanasia, another person administers the intervention intended to cause death. In contemporary jurisdictions permitting the practice, statutory schemes generally concern voluntary euthanasia: a person makes a request meeting legally defined requirements. Non-voluntary euthanasia, involving someone unable to express a valid choice, and involuntary euthanasia, contrary to a person’s wishes, are conceptually different. Authorization of voluntary assisted dying does not imply authorization of these other practices. Historical uses of “euthanasia,” including its use to disguise killings under National Socialism, must also be distinguished from present-day laws requiring voluntary requests.
Canadian medical assistance in dying, commonly abbreviated MAID, includes both practitioner-administered and self-administered forms. Australian voluntary assisted dying statutes also distinguish between these forms, but impose jurisdiction-specific conditions on their use. By contrast, Oregon’s Death with Dignity Act permits a self-administered model rather than clinician-administered euthanasia. The same phrase in an international discussion can therefore describe practices governed by substantially different rules.
Several neighboring practices fall outside these definitions. A competent patient’s refusal of mechanical ventilation or dialysis concerns whether treatment may be continued without consent, not whether a lethal intervention may be provided. Palliative sedation uses medication to reduce awareness when otherwise refractory suffering cannot be adequately relieved; its therapeutic objective is relief of suffering rather than the production of death. The older expression “passive euthanasia” is often avoided because it obscures these distinctions. Intention, consent, proportionality, clinical circumstances, and the immediate cause of death all matter when classifying end-of-life practices.
Historical development and landmark decisions
Modern assisted-dying debates developed alongside advances that made it possible to sustain biological functions despite severe and irreversible illness. Intensive care, resuscitation, long-term ventilation, and artificial nutrition expanded treatment options while creating difficult questions about when treatment could be refused or discontinued. The growth of hospice care and palliative care established another important development: the systematic treatment of pain, breathlessness, psychological distress, and family needs. Assisted dying emerged as a separate legal and ethical question within this changing medical environment.
In the Netherlands, court decisions and professional guidance gradually shaped criteria governing physician conduct before national legislation took effect. The 1973 Postma case became an important landmark in public and legal discussion. The Termination of Life on Request and Assisted Suicide (Review Procedures) Act entered into force in 2002. It established a statutory framework under which physicians who satisfy due-care requirements and report their actions are exempt from criminal liability. Belgium also enacted an euthanasia law in 2002, while Luxembourg followed in 2009. These laws differed in detail despite sharing some underlying concepts.
In the United States, Oregon voters approved the Death with Dignity Act in 1994, and implementation began in 1997 after legal challenges. That year, the United States Supreme Court decided Washington v. Glucksberg and Vacco v. Quill. The Court rejected claims that the federal Constitution established a fundamental right to assistance in suicide or required states to treat assisted suicide and treatment refusal identically. These decisions left substantial room for states to maintain prohibitions or adopt statutory exceptions. In Gonzales v. Oregon in 2006, the Court rejected a federal attempt to use the Controlled Substances Act to prohibit prescribing permitted under Oregon’s law.
Canada’s legal development illustrates the role of constitutional adjudication followed by legislation. In Carter v. Canada in 2015, the Supreme Court of Canada held that the blanket criminal prohibition infringed constitutional rights in specified circumstances involving competent adults who clearly consented and had grievous and irremediable medical conditions causing enduring, intolerable suffering. Federal legislation establishing MAID followed in 2016, with substantial amendments in 2021. These histories show that legalization has occurred through different institutional routes—referendums, parliamentary legislation, and court judgments—rather than through adoption of a single internationally agreed model.
Major legal models and jurisdictional variation
One influential legal model limits assisted dying to adults expected to die within a relatively short period and requires self-administration. Oregon’s statute uses a terminal-illness criterion of a medically confirmed condition expected, within reasonable medical judgment, to cause death within six months. Comparable approaches have informed laws elsewhere in the United States, although procedural requirements vary. A terminal prognosis is a legal eligibility threshold in these systems, not a guarantee about the precise timing of death. Assistance outside the statutory framework remains subject to ordinary criminal and professional law.
The Dutch model does not require a six-month prognosis. Its statutory due-care criteria include a voluntary and well-considered request, unbearable suffering with no prospect of improvement, adequate information, absence of a reasonable alternative, consultation with an independent physician, and medically careful performance. The suffering must have a medical basis. Belgium likewise does not impose a universal six-month requirement, but applies additional procedures where death is not expected in the near future. Neither system permits assistance solely because a person considers life complete without meeting the applicable medical and legal criteria.
Canada’s framework changed significantly in 2021. The requirement that natural death be reasonably foreseeable was removed as a universal eligibility condition, while foreseeability remained relevant to which safeguards applied. The legislation consequently distinguishes between two assessment pathways, commonly called Track 1 and Track 2. The second applies when natural death is not reasonably foreseeable and includes additional safeguards. In 2024, Parliament postponed eligibility where a mental illness is the sole underlying medical condition until 17 March 2027. That was a legislated timetable, not an indication that all other psychiatric eligibility questions were resolved.
Other jurisdictions demonstrate further variation. Switzerland’s criminal law addresses assistance in suicide partly through whether the assistance is motivated by selfish interests; this is not equivalent to a comprehensive medical licensing statute. The person must perform the final act, and professional rules and other laws also apply. Spain’s 2021 law established a regulated euthanasia framework involving advance review. New Zealand’s End of Life Choice Act came into force in 2021 after a referendum. In Australia, Victoria’s scheme began operating in 2019, followed by legislation in other states. Because enactment, commencement, court rulings, and amendments occur at different times, any geographical account must distinguish a law being passed from an assisted-dying service actually becoming available.
Eligibility, capacity, and voluntary choice
Eligibility usually combines medical and decisional requirements. Medical criteria may include terminal illness, an incurable condition, advanced irreversible decline, or suffering that meets a statutory standard. These are not equivalent concepts. A disease can be incurable without being terminal, and a person may have profound functional limitations without experiencing suffering they regard as intolerable. Conversely, severe suffering may occur in conditions with uncertain life expectancy. Legal assessment therefore requires attention to the wording of the governing statute rather than an assumption that a diagnosis alone establishes eligibility.
Decision-making capacity is central to most frameworks. It is generally assessed in relation to a particular decision at a particular time, rather than treated as an all-or-nothing personal attribute. Relevant abilities commonly include understanding information, appreciating how it applies to one’s circumstances, considering available options and consequences, and communicating a choice. A person does not necessarily lack capacity because of age, disability, psychiatric diagnosis, or a decision that others regard as unwise. Conversely, an apparently consistent request does not by itself establish capacity if the person cannot understand its significance.
Voluntariness concerns freedom from coercion and undue influence. Assessment may involve speaking privately with the person, exploring how the request arose, and determining whether someone else is directing or pressuring the decision. Dependence on family or paid caregivers does not itself establish coercion, but it can create circumstances requiring careful evaluation. Financial interests, communication barriers, fear of institutionalization, and statements about being a burden may all be relevant to understanding a request. Statutory witnesses and independent assessments are intended to provide additional checks, though their precise functions differ between jurisdictions.
Informed consent requires more than a signed form. The person must receive and understand relevant information about their condition, prognosis, the nature and consequences of assisted dying, and available alternatives. Prognosis is especially important where eligibility depends on expected survival, but it remains probabilistic. Functional decline, recurrent complications, and response to treatment can alter an estimate over time. A voluntary request also remains revocable: approval does not obligate the person to proceed. Many legal systems require confirmation near the time of provision, subject to narrowly defined exceptions concerning previously documented consent.
Assessment and clinical organization
The clinical pathway generally separates an initial inquiry from a formal request, eligibility assessment, authorization, and possible provision. An inquiry may concern future options rather than a settled intention to proceed. Some people seek information to understand what choices might exist if their condition worsens. A formal request usually activates specific documentation requirements. Clinicians then review the medical record, establish the relevant diagnosis and prognosis, evaluate capacity and voluntariness, and determine whether the person meets statutory criteria. A request may be withdrawn, deferred, found ineligible, or remain incomplete.
Independent assessment is a common safeguard, but its design varies. Many systems require two qualified practitioners; some require particular specialist input or consultation when uncertainty arises. Independence may concern professional relationships, financial interests, or involvement in the person’s treatment. Mental-health consultation is not universally mandatory. In some jurisdictions it is required when impaired judgment or a psychiatric condition affecting capacity is suspected. Specialist expertise may also be needed when the underlying illness is uncommon, the course is unpredictable, or communication difficulties complicate assessment.
Clinical organization differs between self-administered and practitioner-administered models. In the former, a person who receives a prescription may never use it, and the prescribing clinician may not be present when death occurs. In the latter, an authorized practitioner generally confirms the required conditions and is directly involved at the time of provision. These differences affect coordination, documentation, and follow-up. They also influence the completeness of clinical observations: information about the final period may be more detailed when a practitioner is present than when reports depend on relatives or later notification.
Assisted-dying services may be integrated into ordinary clinical care, organized through regional coordination teams, or provided by a relatively small group of participating clinicians. Health-system arrangements determine how assessments are scheduled, records transferred, interpreters obtained, and rural services reached. Conscientious objection adds another organizational dimension. Some laws protect an individual clinician’s refusal to participate while imposing separate duties concerning information, transfer of records, or referral. Institutions may also have participation policies, whose legal status varies. A person’s legal eligibility therefore does not necessarily mean that every practitioner or facility must provide the service, or that access will be equally straightforward in every locality.
Relationship to palliative care and treatment decisions
Palliative care addresses suffering associated with serious illness and can begin well before the final days of life. It includes management of pain, breathlessness, nausea, anxiety, and other symptoms, together with psychological, social, and spiritual support. Its scope is broader than symptom medication alone. Assisted-dying assessments often require discussion of palliative options because a request may arise partly from distress that could be alleviated. However, receiving palliative care and requesting assisted dying are not mutually exclusive; administrative reports show that many recipients in jurisdictions permitting assisted dying have also received palliative services.
The relationship between symptom relief and requests is not simple. Physical symptoms may be important, but reported concerns also include loss of independence, inability to engage in valued activities, reduced control over bodily functions, and fear of future deterioration. Such concerns can persist despite effective treatment of pain. They can also change with improved support, rehabilitation, communication assistance, or resolution of a crisis. A clinical account must therefore distinguish currently uncontrolled symptoms from anticipated suffering and from broader judgments about what makes continued life acceptable to the individual.
Palliative sedation is a separate intervention for refractory suffering. Depending on the circumstances, sedation can vary in depth and duration; continuous deep sedation until death is a specific practice rather than a synonym for all sedative use near death. Professional frameworks emphasize careful assessment, proportionality, consent, and documentation. The distinction from assisted dying concerns the intervention’s aim and method, not simply whether death occurs afterward. Similarly, appropriately titrated opioid treatment for pain or breathlessness should not be classified as assisted dying merely because the patient is close to death.
Withholding or withdrawing treatment raises another set of questions. A person may decline cardiopulmonary resuscitation, ventilation, artificial nutrition, or other interventions, subject to the applicable law. The resulting death is ordinarily attributed to the underlying condition rather than to an act intended to cause death. Decisions about hydration and nutrition require particular attention to whether support constitutes medical treatment and to the circumstances of the patient. The doctrine of double effect is sometimes invoked in end-of-life ethics to distinguish intended therapeutic benefit from foreseeable but unintended harm, but it does not replace clinical evidence, consent requirements, or jurisdiction-specific legal rules.
Psychiatric illness, cognitive impairment, and communication
Psychiatric conditions present several distinct questions that should not be conflated. A person with a terminal physical illness may also have depression or anxiety; these conditions may or may not affect capacity or the voluntariness of a request. Separately, some legal systems permit consideration of assisted dying when a psychiatric disorder itself is the principal basis of suffering. Many others exclude that possibility. The presence of a psychiatric diagnosis does not automatically invalidate a decision, but severe hopelessness, delusional beliefs, or impaired appreciation of treatment possibilities can be relevant to assessment.
Where psychiatric conditions can qualify, evaluating irremediability is particularly complex. Unlike some advanced malignancies, psychiatric disorders often have fluctuating courses and no single biological measure that establishes a fixed prognosis. Treatment history, duration, previous responses, comorbidity, and the availability of further interventions may all matter. A judgment that suffering is intolerable is also different from a judgment that no reasonable prospect of improvement remains. Dutch professional guidance and review practice apply particular caution to psychiatric requests, including specialist input. Such requirements reflect the specific evidential difficulties rather than an assumption that all people with mental disorders lack capacity.
Dementia raises the problem of a progressive loss of decision-making ability. A person may retain capacity early in the disease but lose it before reaching a stage previously described as unacceptable. An advance directive can record future treatment preferences, but authority to refuse treatment through a directive does not automatically authorize assisted dying. Some jurisdictions allow a written euthanasia request to have legal significance after capacity is lost, subject to additional requirements; others require contemporaneous consent and therefore exclude this pathway. Difficult cases concern the interpretation of earlier wishes, present behavior, and evidence of current suffering.
Communication impairment must also be distinguished from cognitive impairment. Aphasia, severe motor disability, or loss of speech can make conventional interviews difficult without eliminating understanding or choice. Interpreters, accessible materials, assistive technology, and specialist communication assessments may help establish what a person understands and intends. Legal requirements concerning who performs the final act can create an additional distinction: a person may have full capacity yet be physically unable to self-administer. Whether clinician administration is then available depends on the jurisdiction’s law, not solely on a clinical judgment about the person’s ability to consent.
Ethical, professional, and human-rights questions
Ethical discussion commonly draws on autonomy, beneficence, non-maleficence, and justice. Autonomy concerns a person’s authority over decisions affecting their body and life. Beneficence concerns relief of suffering, while non-maleficence concerns avoiding harm. Justice concerns fair treatment, protection, and the distribution of health and social resources. These principles do not mechanically determine a single conclusion. Disagreement concerns their interpretation and relative weight: for example, whether deliberately causing death can form part of medical care, and whether respecting a voluntary request can justify professional participation.
Professional organizations adopt differing positions. The World Medical Association’s 2019 declaration maintained opposition to euthanasia and physician-assisted suicide while distinguishing them from respect for treatment refusal and appropriate end-of-life care. Other national organizations have adopted neutral positions or developed guidance for members working within laws permitting assisted dying. A neutral organizational position on legislation does not necessarily mean that the organization regards all forms of practice as clinically equivalent. Professional guidance can address competence, conflicts of interest, documentation, and continuity of care without requiring every member to participate.
Human-rights law has also produced context-specific judgments rather than a universal legal entitlement. In Pretty v. United Kingdom in 2002, the European Court of Human Rights held that the right to life under Article 2 of the European Convention could not be interpreted as conferring an opposite right to die. The Court’s jurisprudence has nevertheless recognized that questions about how and when life ends can engage respect for private life under Article 8. States retain a margin of appreciation in regulating the field. Canada’s Carter decision arose under a different constitutional framework and cannot simply be transferred to other legal systems.
Disability-rights discussion focuses particularly on equality and the social conditions surrounding choice. Disability organizations and advocates hold differing positions, but recurring concerns include inaccessible care, inadequate personal assistance, poverty, and assumptions that dependence necessarily makes life intolerable. These concerns distinguish an individual’s expressed wishes from social judgments about the worth of disabled lives. Conversely, arguments concerning equal decisional authority emphasize that disability alone should not remove a competent adult’s agency. The central policy question is how to evaluate an individual request without either presuming incapacity or overlooking the influence of remediable disadvantage.
Reporting, oversight, and interpretation of statistics
Oversight systems operate at different points in the process. Some require advance authorization by a commission or other body; others rely primarily on practitioner assessment followed by retrospective review. Dutch regional euthanasia review committees assess reported cases against statutory due-care criteria. Belgium has a federal control and evaluation commission. In the United States, state health departments commonly collect prescribed information under assisted-dying statutes, but their powers and review functions differ from those of European review bodies. A reporting system is therefore not necessarily an independent authorization system.
Published figures measure different events. A request, an eligibility assessment, a prescription, and a death following administration are separate observations. Some people die from their illness during assessment, withdraw their request, or receive medication without taking it. For self-administered systems, deaths in a reporting year may involve prescriptions issued in an earlier year. Cross-jurisdiction comparisons become misleading if one system reports prescriptions while another reports completed cases. The denominator also matters: an assisted-dying count expressed as a proportion of all registered deaths is not the same measure as a population rate.
For 2023, Health Canada reported 15,343 MAID provisions, representing approximately 4.7% of deaths in Canada. The Netherlands’ regional review committees reported 9,068 notifications for 2023, approximately 5.4% of deaths. These figures describe different statutory populations and reporting arrangements; they do not establish that the systems have equivalent eligibility criteria or levels of access. Counts can also change because of population aging, service availability, professional participation, legislative amendments, or reporting practices. A numerical increase alone does not identify which mechanism is responsible.
Medical certification creates further variation. Jurisdictions differ in how assisted dying, the underlying disease, and the manner of death are recorded on a death certificate. Aggregate mortality data may consequently be insufficient to identify assisted-dying cases without a dedicated reporting system. Confidentiality rules can also limit publication of small subgroups. Review findings require similar care: a determination that reported cases met formal criteria does not by itself establish that every potentially relevant circumstance was detected, while an identified procedural breach does not necessarily show that the person was medically ineligible. Administrative oversight and population-level research answer related but distinct questions.
Evidence, access, and unresolved research questions
The evidence base consists largely of statutory reports, medical-record studies, surveys, interviews, and observational comparisons. These sources illuminate who requests assisted dying, which illnesses are recorded, how clinicians conduct assessments, and how families experience the process. They are less able to establish what would have happened to the same individuals under a different legal or care system. Cohort studies can follow applicants or recipients over time, but selection into requesting, receiving, and participating in research creates substantial interpretive problems. Bereaved-family studies additionally depend on who agrees to be interviewed and when interviews occur.
Measures of suffering and motivation require particular caution. Administrative forms often record several reasons simultaneously, and categories such as loss of autonomy or perceived burden can cover very different experiences. Clinician-recorded reasons may not fully match the person’s own account. Patient-reported outcomes can add detail, but serious illness, fatigue, communication impairment, and death limit repeated measurement. A request itself may evolve over time: an initial wish to explore an option, a sustained formal request, and a decision to proceed should not be treated as identical outcomes.
Access is shaped by geography, language, income, institutional policy, and the availability of willing and qualified practitioners. These influences intersect with social determinants of health and health equity. A low rate of use in a population may reflect preference, lack of awareness, legal ineligibility, barriers to assessment, or cultural attitudes; it cannot be interpreted as evidence of either protection or exclusion without further information. Similarly, high uptake does not on its own establish that care is more responsive or that safeguards are weaker. Research must distinguish barriers to a lawful assessment from differences in the underlying frequency of requests.
Several recurring claims remain difficult to test causally. These include effects on trust in medicine, non-assisted suicide, palliative-care development, professional behavior, and experiences of socially disadvantaged groups. Comparisons before and after legalization are affected by changing demographics, health services, reporting systems, and broader social conditions. Studies of safeguard performance also face the difficulty that undetected coercion or unreported practice is, by definition, incompletely observed. More comparable definitions, transparent reporting of incomplete and declined requests, and careful separation of association from causation are central methodological issues in this field.