Comparative Law Recent Things 139 Harv. L. Rev. 1984

Barriers and Safeguards: Access to Assisted Dying for the Socioeconomically Vulnerable


Download

Introduction

The current scope of global access to assisted dying is unprecedented. It is also expanding. Assisted dying is legal in fifteen countries and parts of two others,1 and at least four other countries are actively considering legalization.2 The vast majority of jurisdictions with assisted dying frameworks decriminalized or legalized the practice within the past fifteen years, and 37.5% did so within the past five years.3 Within this wave of new legislation roil debates about how assisted dying frameworks do and should affect vulnerable individuals. Some advocacy groups cite specific risks to the autonomy of vulnerable people, maintaining that “no number of safeguards . . . could be put in place that [would] offset the pressure” on vulnerable individuals to access assisted dying services.4

There are myriad ways to structure legal frameworks that govern assisted dying. Whatever their forms and particulars, they ought to be attentive to the unique needs of vulnerable individuals. The UK House of Commons’ Public Bill Committee recently considered, extensively debated, and rejected the possibility of excluding prisoners and unhoused people from accessing assisted dying.5 Several members criticized the proposed exclusion as a mechanism for discrimination,6 while another expressed concern that omitting the exclusion would create opportunities for vulnerable individuals to be improperly pressured into seeking assisted deaths.7 But neither these concerns nor the legal measures taken to address them have been robustly tested.

It can be difficult to discern whether carefully constructed legislative strategies actually protect the interests of vulnerable people as intended. Recent analyses suggest that people with relatively low socioeconomic status are less likely to access assisted dying services, across international jurisdictions.8 There are probably a host of reasons behind that finding: People with lower socioeconomic status tend to have reduced access to health care services generally9 — and particularly end-of-life care10 — as a function of rural geography, difficulty navigating administrative complexities, health care costs, and ancillary costs like transportation.11 Another contributing factor could be how lawmakers choose to structure frameworks for assisted dying administration. If the systems we’ve designed to protect vulnerable patients actually exclude them, as UK legislators suggest they might, is there a misalignment between the actual and intended effects of existing assisted dying safeguards?12

This Note makes two contributions: First, Part II summarizes existing literature and supplies novel statistical analyses to reveal that, even in jurisdictions with supposedly permissive assisted dying frameworks, there’s no conclusive evidence of an association between socioeconomic vulnerability and an increased likelihood of accessing assisted dying. Second, Part III compares all international assisted dying frameworks across twelve axes, providing a comprehensive matrix of the restrictions that judges and lawmakers have imposed on assisted dying access. The first contribution undermines an established ethical critique of assisted dying legalization — which is outlined at Part III — by challenging the idea that assisted dying poses special risks to socioeconomically vulnerable people. The second contribution offers an infrastructure for scholars and lawmakers searching for ways to protect the interests of socioeconomically vulnerable patients who seek assisted dying services, and to assess the efficacy of those protective measures.

Before proceeding, a brief note on terminology is in order: This Note uses the term assisted dying to describe all methods by which a person acts to hasten the death of another person, at the recipient’s voluntary request.13 Different jurisdictions and scholars use various labels to describe assisted dying practices.14 The analysis below adopts the following definitions: Assisted dying encompasses both physician-assisted dying, where a health care provider helps a person to terminate their own life by providing medication for self-administration, and forms of assisted dying where a person (generally a health care provider15) administers medication to end the life of the recipient.16

I.  Vulnerability and Autonomy Discourse: The Background Inequality Critique

There are many ways to define and measure vulnerability, from both legal and bioethical perspectives.17 Some authors frame the concept by identifying specific at-risk groups whose members are more likely to experience undesirable outcomes, like coercion.18 Professor Martha Fineman, on the other hand, emphasizes universality: Each of us will be vulnerable at some point in our lives, by virtue of age, illness, injury, or social factors, so it’s a mistake to define vulnerability in binary at-risk/not-at-risk terms.19

This Note focuses on the socioeconomic facet of vulnerability, which has become a critical part of discussions about assisted dying.20 When the U.S. Supreme Court heard Vacco v. Quill21 in 1997, amici curiae from across the country wrote to warn the Court of the danger assisted dying might pose to impoverished people.22 Individual cases continue to inspire (often heavily sensationalized23) headlines.24 Debating the UK’s recent assisted dying bill, Member of Parliament Lewis Atkinson said the following about the nexus of socioeconomic vulnerability and end-of-life decision making: “At the moment, at the end of life, wealth clearly is advantaged. Those who are wealthy and are able to go to Switzerland do have choice at the end of life. Dignity and independence and autonomy should not be based on ability to pay.”25

Atkinson clips the surface of a rich debate. Among the arguments relevant to that debate is the background inequality critique, which proceeds in two steps: First, socioeconomically vulnerable individuals might choose assisted dying as a result of their vulnerability26 — for example, because they feel like burdens on their support networks27 or socioeconomic suffering leaves them unable to access supportive care.28 Even if these individuals want to access assisted dying, that preference may therefore be “adaptive” or “inauthentic.”29 Second, no number or arrangement of safeguards would be sufficient to alleviate this concern, insofar as background social inequalities in access to palliative care and other resources continue to effect pressure on the decisions of vulnerable people.30

The background inequality critique doesn’t address socioeconomic vulnerability alone, and this Note’s focus on socioeconomic aspects of vulnerability shouldn’t be read to imply as much. Proponents highlight meaningful intersections between poverty and disability, articulating a concern that people will seek assisted deaths not only as a direct result of poverty,31 but also because of the way poverty confounds proper treatment of their existing health conditions.32 More surreptitiously, poverty across an individual’s lifespan can make them more likely to develop health conditions, and specifically poor prognoses stemming from those health conditions, that could underlie a decision to seek assisted dying.33

This critique is also not an ethical slippery slope argument. Proponents of slippery slope arguments posit that legal frameworks seeming at first blush to be sufficiently protective of vulnerable people will necessarily engender “unwanted, unintended, and undesirable events,”34 like extending assisted dying to certain minors.35 But according to the background inequality critique, the proverbial slope has no peak — the argument is not that one regulation will spur the promulgation of other, less desirable ones, but that no conceivable set of restrictions can produce a just system of assisted dying administration.36

Scholars have offered compelling counterarguments to the background inequality critique. Adapting language that Justice Lynn Smith had used in an appellate decision37 that “paved the way” for the Supreme Court of Canada to announce the decriminalization of assisted dying in Carter v. Canada (Attorney General),38 Professor Udo Schüklenk argues that the critique “hold[s] decisionally capable people . . . hostage to the establishment of [a] perfect system” that will almost certainly never materialize.39 And the critique doesn’t provide much guidance as to how harms should be mitigated after legalization. Professors Alexandra Mullock and Jonathan Lewis argue that banning assisted dying out of concern for the autonomy of vulnerable individuals can have the paradoxical effect of creating “pathogenic vulnerability”40 because it presents new threats to autonomy, like being unable to afford travel to a permissive jurisdiction.41

One question central to ethical conversations about assisted dying is therefore whether its recipients tend to have, empirically speaking, relatively limited financial means. The literature review and analyses at Part II suggest that they do not; this undercuts the background inequality critique because it provides descriptive evidence that patients may be choosing, in the statistical aggregate, to access assisted dying for reasons other than socioeconomic vulnerability or its effect on their physical and mental health.

II.  Documenting and Interpreting Rates of Assisted Dying Uptake by Socioeconomically Vulnerable Patients

Socioeconomic vulnerability, even apart from its intersections with other forms of vulnerability, can be defined malleably and expansively. One common strategy is to aggregate various factors contributing to socioeconomic status,42 like financial stability, educational attainment, housing security, and employment.43 This descriptive plurality also pervades the ways that international jurisdictions choose to structure centralized reporting on assisted dying. Almost every jurisdiction that has decriminalized or legalized assisted dying maintains a reporting system and issues annual descriptive analyses,44 but they rarely agree on precisely which data should be reported.45 Data collection is particularly scattershot with respect to socioeconomic indicators.46

To investigate the impact of socioeconomic vulnerability on assisted dying access is therefore to navigate a thicket of proxies. No local authority has directly linked assisted dying data to household income data. But, as discussions about socioeconomic vulnerability plainly contemplate more than just gross income levels,47 so too should statistical analyses of socioeconomic vulnerability. An individual’s employment status, education history, insurance coverage, and location of residence can also tell us something about whether they are likely to live on higher or lower incomes, or to have done so in the past, and how resilient they are against the economic pressures of a serious illness.48

This Part divides publicly reported descriptive analyses into two subject matter categories — income data and education data — to summarize available information that can be used to assess how socioeconomically vulnerable people use assisted dying services. Neither the results of independent studies nor novel analyses of government summary reports support the contention that socioeconomic vulnerability makes affected individuals more likely to choose an assisted death.

A.  Geographically Estimated Income Data

No jurisdiction in the world directly collects data respecting the personal incomes of assisted dying recipients,49 so scholars and governments have used patients’ residential addresses to estimate their income quintiles. The literature boasts five comparisons of estimated incomes among assisted dying recipients to income levels in a reference population — four from Canada50 and one from Switzerland.51 None revealed that individuals with relatively low estimated incomes were disproportionately likely to access assisted dying.52 A few other studies have assessed socioeconomic distribution among assisted dying recipients, but without comparison to a broad reference population.53 Provincial and federal governments in Canada have also used marginalization indices to describe the socioeconomic statuses of assisted dying recipients as opposed to those of people in a baseline population — with mixed results.54

B.  Education Data

Formal education is a useful indicator of socioeconomic vulnerability because it correlates with higher incomes.55 U.S. states consistently document the educational attainment level of assisted dying recipients,56 while other jurisdictions don’t report educational data57 or do so inconsistently.58 And while other scholars have aggregated these data in the past, they don’t situate their analyses using a baseline reference population.59 This omission is meaningful because the proportions of assisted dying recipients with certain educational attainment levels tell us precious little unless we know what we should expect to observe. Assume, for example, that 25% of people who access assisted dying don’t hold high school diplomas — this value could be disproportionately low or high, depending on the percentage of people in the baseline population who hold diplomas.

One limitation affecting U.S. analyses is a lack of public access to information about individual patients (or microdata). While other jurisdictions maintain microdata sets and carefully restrict access,60 every U.S. state that provides for centralized reporting prohibits disclosure of microdata to any member of the public.61 California’s statute forbids disclosure even to a court of law.62 This poses a transparency issue because it heavily restricts analysis of whether U.S. assisted dying frameworks are being implemented safely, fairly, and equitably.

Nonetheless, summary data offered by various U.S. states can facilitate meaningful comparisons between assisted dying recipients and reference populations. Table 1 shows the most recent four years of assisted dying and educational attainment data across states who report it and compares these values to state-specific estimates of educational attainment from the Census Bureau’s 2023 American Community Survey.63

The analysis at Table 1 is limited in the following ways: First, it uses census data as a reference, which is generally useful in the U.S. context because all but three states limit access to permanent in-state residents.64 Data from one of these three states suggest that very few nonresidents use the service.65 It would have been preferable to use a more narrowly tailored reference population, such as data respecting everyone who requested access (which is unavailable because U.S. states don’t report data for requests that were ruled ineligible66), everyone who was eligible to make a request (which is unavailable because eligibility is assessed only for those who make requests67), or everyone who died of any cause within the reporting year (which is unavailable because U.S. states don’t publicly report educational data for this group68). Second, educational attainment is an imperfect proxy for income. While individuals with relatively high educational attainment earn more, on average, than those without,69 the financial circumstances of specific individuals may vary. (A tenured professor with a doctorate, for example, could become impoverished because illness left her unable to work.) And third, the analysis is confounded by age. While some of the people within the samples are too young to have earned postsecondary degrees, they must be included because it isn’t possible to stratify assisted dying educational data by age without information about individualized entries.70 As a robustness check pertaining to age, Table 1 also provides a reference value for people aged sixty-five years and up to facilitate assessment of whether older adults (who are more likely to suffer from terminal illnesses71) are disproportionately more educated than their younger counterparts. If this were true (which Table 1 reveals is not the case72), using a reference population of all adults could have overestimated the expected proportion of assisted dying recipients with relatively little formal education because recipients are older, on average, than the reference group.73

Table 1: Educational Data

Reference Population74

Assisted Dying Data75

# of cases (% of annual cases)

Aged 18+ yrs

Aged 65+ yrs

2021

2022

2023

2024

California76

n=30,527,864

HS/GED/Unreported77

36.8%

121 (23.2%)

201 (23.6%)

222 (22.9%)

256 (24.8%)

Some/AD78

28.6%

121 (23.2%)

210 (24.6%)

252 (26.0%)

262 (25.4%)

Bachelor’s/Higher79

34.6%

33.8%

280 (53.6%)

442 (51.8%)

495 (51.1%)

514 (49.8%)

Colorado80

n=4,666,930

HS/GED/Unreported

28.4%

43 (21.0%)

58 (19.7%)

85 (23.5%)

116 (27.0%)

Some/AD

28.8%

39 (19.0%)

58 (19.7%)

80 (22.1%)

94 (21.9%)

Bachelor’s/Higher

42.8%

43.1%

123 (60.0%)

179 (60.7%)

197 (54.4%)

219 (51.0%)

District of Columbia81

n=552,446

HS/GED/Unreported

21.6%

1 (16.7%)

0

Unpublished

Some/AD

16.8%

0

0

Bachelor’s/Higher

61.6%

49.7%

5 (83.3%)

8 (100%)

Hawaii82

n=1,141,251

HS/GED/Unreported

35.2%

16 (55.2%) (12 unreported)

18 (48.6%) (10 unreported)

26 (51.0%) (23 unreported)

Unreported

Some/AD

30.5%

1 (3.4%)

3 (8.1%)

5 (9.8%)

Bachelor’s/Higher

34.3%

35.1%

12 (41.4%)

16 (43.2%)

20 (39.2%)

New Jersey83

n=7,281,676

HS/GED/Unreported

35.4%

11 (22.0%)

33 (36.3%)

Excluded84

41 (33.6%)

Some/AD

23.3%

6 (12.0%)

5 (5.5%)

17 (13.9%)

Bachelor’s/Higher

41.3%

35.0%

33 (66.0%)

53 (58.2%)

64 (52.5%)

Oregon85

n=3,403,557

HS/GED/Unreported

32.2%

72 (28.2%)

73 (26.3%)

112 (29.0%)

122 (32.4%)

Some/AD

32.8%

66 (25.9%)

69 (24.8%)

95 (24.6%)

86 (22.9%)

Bachelor’s/Higher

35.0%

34.5%

117 (45.9%)

136 (48.9%)

179 (46.4%)

168 (44.7%)

To provide conservative estimates, Table 1 adds unreported entries to the “High School Diploma, GED, or Less” category. This mitigates any concern that unreported entries could be concealing the true proportion of recipients within this category because patients and physicians are, for some reason, less likely to document education levels for patients with lower levels of attainment. This strategy also has the effect of skewing results in favor of finding disproportionate access by individuals with relatively low educational attainment.

Across all four reporting years, there are five instances where the proportion of assisted dying recipients within the “High School Diploma, GED, or Less, or Unreported” attainment level exceeds the corresponding reference proportion. According to careful analysis using hypothesis tests of proportions, none of these five results establishes that people occupying these (relatively low) educational attainment categories are accessing assisted dying at a disproportionate frequency.

1.  New Jersey. — One of the five instances arose in New Jersey during the 2022 reporting period, where the proportion of recipients in the lowest educational attainment category was 0.9 percentage points higher than the reference proportion. This Note uses a one-sided, one-sample test of proportions based on the binomial distribution to assess whether this deviation from the reference proportion was statistically significant, using a standard significance level of 5%.86

Assuming the null hypothesis is true, there was a 47.1% probability of observing thirty-three or more individuals within the “High School Diploma, GED, or Less, or Unreported” category, as New Jersey officials did in 2022.87 Since this p-value exceeds the 5% threshold, these data do not justify the rejection of the null hypothesis that the relatively high proportion in 2022 can be attributed to random chance.

2.  Oregon. — In Oregon’s 2024 reporting year, the proportion of assisted dying recipients within the lowest attainment category exceeded the baseline proportion by 0.2 percentage points. Applying the same hypothesis test of proportions as in subsection III.B.1 above,88 the resulting p-value is 0.479. This result likewise does not justify the rejection of the null hypothesis.

3.  Hawaii. — The remaining three highlighted instances arose in Hawaii during the 2021, 2022, and 2023 reporting years. Applying the same hypothesis test produces the following results89: p-values of 0.022 for the 2021 reporting year,90 0.064 for 2022,91 and 0.015 for 2023.92 At first glance, these results indicate that we should reject the null hypothesis as to 2021 and 2023 — in other words, that the elevated representation of individuals within the lowest attainment category during these reporting years cannot be attributed to random chance alone.

But per Table 1, nonreporting was unusually prevalent in Hawaii, where up to 45.1% of entries (in 2023) did not include educational data. Table 1 assumes, for the sake of caution, that every unreported entry was a recipient in the lowest attainment category. But if one excludes all unreported entries from the analysis, access by individuals falling into the lowest category doesn’t exceed the reference proportion in any of the three relevant reporting years: As opposed to 35.2% in the reference population, 23.5% of recipients from the 2021 sample,93 29.6% from the 2022 sample,94 and 10.7% from the 2023 sample95 fall within the lowest category. The disappearance persists if one imputes unreported entries according to the proportions observed in reported entries, such that 24.1% of 2021 cases,96 29.7% of 2022 cases,97 and 9.8% of 2023 cases98 fall into the lowest category.

4.  Low Educational Attainment Is Associated with Lower Uptake of Assisted Dying Services. — To summarize, the four most recent years of data respecting the educational attainment characteristics of assisted dying recipients don’t indicate that people with relatively low educational attainment are accessing assisted dying with greater frequency than those with relatively high attainment. In fact, hypothesis testing at Table 2 reveals that the opposite is true: People with relatively low educational attainment are disproportionately unlikely to access assisted dying services, and this difference is statistically significant. Table 2 applies the same methodology described at subsections III.B.1 to III.B.3 to every instance where the number of assisted dying recipients in the lowest educational attainment category fell below the reference proportion.

Table 2: Low Educational Attainment Is Associatedwith Reduced Uptake of Assisted Dying99

2021

2022

2023

2024

California

< 0.0001***

< 0.0001***

< 0.0001***

< 0.0001***

Colorado

0.0098***

0.0004***

0.0204**

0.286

Hawaii100

0.146

0.305

< 0.0001***

Unreported

New Jersey

0.0301**

Higher than reference proportion

Excluded from Table 1

0.378

Oregon

0.0979*

0.0186**

0.0821*

Higher than reference proportion

P (X≤x)

x=number of people in the lowest educational attainment category

in a given year

P=probability of observing the “High School Diploma, GED, or Less,or Unreported” result within the census baseline

n=total number of assisted dying recipients in the same year

Stars indicate significance level: *** p < 0.01; ** p < 0.05; * p < 0.1

Thus, on most occasions when assisted dying uptake by individuals in the lowest educational attainment category was lower than the reference proportion, the magnitude of that difference was not attributable to random chance. One possible explanation for this result is that U.S. frameworks impose barriers to assisted dying access that are disproportionately burdensome to individuals living on relatively low incomes.

Table 2’s results — as well as existing efforts to assess the socioeconomic distribution of assisted dying recipients101 — therefore reveal no trend toward disproportionate access by people with relatively low socioeconomic status. This finding both undermines the background inequality critique’s premise that socioeconomic vulnerability (among other forms of social vulnerability) is likely to drive people to choose assisted dying, and also raises concerns that assisted dying frameworks may be disproportionately excluding socioeconomically vulnerable people as a side effect of various measures to safeguard access.102

But this conclusion merits a few important qualifications: First, in spatially large jurisdictions with vast disparities in urban and rural health care access, the socioeconomic distribution of assisted dying uptake may look different as to urban or rural residents than it does in the aggregate — so optimally, reference data would be specific to each region. But data from one state suggest that rural residents make up a very small proportion of assisted dying recipients,103 so these disparities are difficult to assess and unlikely to make a meaningful analytical difference. Second, variability in jurisdictions’ data collection practices means that direct comparisons aren’t always possible. Third, it’s often impossible to resolve this comparability issue because many jurisdictions don’t allow public access to assisted dying microdata. U.S. states, for example, are bound by statute to withhold this data from the public.104 And fourth, descriptive analyses like this one can show how many people from a specific socioeconomic group accessed assisted dying, but cannot assess confounding demographic factors — like the prevalence of terminal underlying diagnoses due to poor preventive care access — that could help to explain those results. Available data are therefore by no means irrefutably conclusive on the question of how legislative decisionmaking impacts the probability that socioeconomically vulnerable people will access assisted dying.

III.  A Comprehensive Descriptive Matrix for the Restrictiveness of International Assisted Dying Programs

The empirical analyses at Part II supply a partial answer to the background inequality critique. They suggest that, at least in the statistical aggregate, socioeconomically vulnerable people are not rushing to access assisted dying services more frequently than their relatively less vulnerable counterparts. But a second and equally critical question is whether specific individuals, apart from the statistical whole, are adequately protected from coercion by existing safeguards. While “[p]overty is not incompatible with autonomous decision-making,”105 “economically disadvantaged individuals do in fact make decisions differently as a result of their experience of poverty.”106 The contention that poverty is not itself a ground to withhold assisted dying access should not be confused for a declaration that no one can exert financial pressure against a patient,107 or that remediation of suffering — including suffering exacerbated by financial scarcity — needn’t be prioritized.

Though the background inequality critique doesn’t reflect statistical reality, it may yet encourage scholars and legislators to think carefully about how assisted dying delivery frameworks can best respect the needs of vulnerable individuals. This Part does more to contextualize the background inequality debate by articulating the ways that legal decisionmakers have tried to address the potentially coercive effect of socioeconomic vulnerability. It also sets the stage for future inquiries into whether these safeguarding measures are up to that task — or perhaps overreaching it, as UK lawmakers feared.108

There’s considerable uniformity of thought among jurisdictions with comprehensive assisted dying legislation. In Germany,109 Italy,110 and Montana,111 constitutional courts have overturned criminal prohibitions on assisted dying, but legislators have yet to implement frameworks for access, and in Switzerland, legislators decriminalized assisted dying in 1937112 but never enacted a comprehensive framework. In all other relevant jurisdictions, the following requirements hold fast: Requests must be voluntary and enduring, in that they’re repeated over time and may be withdrawn at any time;113 recipients must have an underlying illness or condition;114 and health care providers must inform recipients about their illnesses or conditions, the alternatives available to them, and the assisted dying procedure.115

But in other respects, legislative approaches differ widely. Required waiting periods between initial and final requests range from ninety days to twenty-four hours, and may or may not be waivable.116 Some jurisdictions allow mental illnesses to satisfy diagnostic requirements for accessing assisted dying, while others specifically foreclose that possibility,117 treat individuals experiencing mental illnesses with elevated scrutiny,118 or leave this question entirely open.119 Commenters often refer qualitatively to the restrictiveness of assisted dying frameworks,120 but to date have taken few pains to compare different frameworks in terms of the safeguards they impose. As a result, different voices use permissive, restrictive, and similarly polar terminology to describe the same frameworks: Commenters have described the Netherlands’s decades-old framework as both “really strict”121 and “worryingly permissive.”122 This creates difficulties for scholars and legislators by obscuring how different legislative safeguards have actually affected vulnerable patients and how these effects should inform future lawmaking.

In response to this confusion, this Note proposes a comprehensive descriptive matrix — detailed in full at the Appendix123 — that outlines each jurisdiction’s assisted dying framework in twelve categorical respects. Each of these twelve restrictions has been selected because evidence indicates that it curtails access, procedurally and/or substantively, for a meaningful number of people who might otherwise use assisted dying services. In other words, this Note defines “restrictiveness” as a measure of an assisted dying framework’s capacity for exclusion of particular groups of people from access; the more categorical restrictions a framework imposes, the more restrictive it is.

It’s important to note that a relatively low number of “restrictions” doesn’t necessarily make it logistically easier to access assisted dying. For example, Colombia’s Ministry of Health and Social Protection amended assisted dying regulations in 2021 to extend eligibility to patients lacking terminal diagnoses,124 but six months would pass before the first patient without a terminal illness accessed assisted dying, amid much debate and frustration.125 In the Netherlands, a shortage of psychiatrists means that it takes an average of two years for a patient to access assisted dying for suffering stemming from a mental illness.126 So legislative restrictions on who may access assisted dying and how eligible persons may do so are just one set of barriers, or safeguards, to access.

To describe assisted dying frameworks in terms of their restrictiveness, this Note identifies a comprehensive scheme of twelve constraints on access that policymakers have incorporated to date: (1) prohibiting anyone other than the patient from administering assisted dying medication;127 (2) requiring that death be expected within a specific timeframe;128 (3) barring the use of mental illness as a diagnosis justifying a request;129 (4) prohibiting the use of advance directives for individuals who lose decisionmaking capacity;130 (5) requiring that patients satisfy certain diagnostic criteria, like incurability;131 (6) excluding patients who are minors;132 (7) imposing a citizenship or residency requirement;133 (8) requiring confirmation of eligibility from two health care providers;134 (9) requiring that a committee confirm eligibility;135 (10) requiring a written declaration of intent from the patient;136 (11) imposing a waiting period between an initial request and receipt of assisted dying;137 and (12) enforcing centralized monitoring and reporting.138

This comprehensive matrix yields surprising results. For example, some sources characterize Canada’s requirements as “among the broadest in the world”139 and “looser than those of Belgium and the Netherlands,”140 but this claim doesn’t bear out empirically. Among seventeen countries that have decriminalized or legalized assisted dying across the nation or in specific states, eleven impose fewer categorical restrictions than Canada or contain at least one state that does.141 Even excluding countries where the practice is decriminalized by way of a constitutional court decision without any legislative framework,142 Canada surfs the middle of the restrictiveness wave. A similar phenomenon afflicts Belgium’s framework, which commenters have erroneously called “the world’s most liberal euthanasia law.”143 The matrix also clarifies various similarities and distinctions among international frameworks. Centralized reporting, for example, is a nearly universal restriction, whereas only four jurisdictions allow mental illness to be a ground for an assisted dying request144 and roughly half impose residency or citizenship requirements.145

Sometimes, generalizations about the restrictiveness of an assisted dying framework may reveal more about whether an author approves of a particular component thereof than the comprehensive picture of relevant law within the jurisdiction. For example, one source argues that Canada originally had “strict eligibility criteria” but is now exceedingly permissive, and attributes that change to the legislature’s 2021 decision to remove a requirement that death be “reasonably foreseeable.”146 And the lack of a categorical restriction against nonterminal diagnoses is a common subject of discourse on so-called “suicide tourism” in Switzerland,147 though it’s only one factor contributing to the overall restrictiveness of an assisted dying framework.

Objective, systematic descriptions of assisted dying frameworks are therefore critical to facilitating productive discussion about how legislation can serve the particularized needs of socioeconomically vulnerable people seeking assisted death. While any attempt to describe the restrictiveness of legislation is vulnerable to potential bias — for example, stemming from the selection of the categories — a descriptive matrix mitigates (or at least elucidates) those biases by forcing them into the open.

Another important use of the matrix is to situate future comparisons of international data respecting assisted dying uptake. At present, no one has attempted a direct comparison of international data on the socioeconomics of assisted dying access, and indirect comparisons are necessarily imperfect. But if we do observe differences in socioeconomic indicators as between different jurisdictions, we can assess whether those differences correlate with specific access restrictions. For example, both existing Swiss research and the section II.B U.S. education analyses revealed a statistically significant correlation between higher (estimated) socioeconomic status and the probability of accessing assisted dying.148 So this correlation seems to hold even among countries that impose vastly different legislative safeguards on assisted dying access.

This descriptive scheme should be qualified twice over. First, its author was responsible for selecting the categorical restrictions. This selection process was not arbitrary, but it was necessarily vulnerable to bias — to mitigate this concern, this Note provides a detailed explanation of the justifications for including and excluding specific variables. Second, the scheme doesn’t address extralegal constraints on the delivery of assisted dying services, like professional guidelines. While some professional bodies may ground disciplinary decisions on noncompliance with professional guidelines, these guidelines don’t carry the coercive force of law. Also, the restrictiveness scheme is intended in part to guide legislative decisionmaking, and lawmakers cannot predict before enacting or amending a statute which guidelines professional organizations might impose.

Conclusion

As a statistical matter, available data indicate that people with indicators of socioeconomic vulnerability aren’t accessing assisted dying at an outsized frequency; this distinctly undermines the background inequality critique of assisted dying legalization. But this doesn’t mean that any given individual is impervious to the risk of coercive pressures stemming from a lack of financial resources. Understanding existing restrictions on assisted dying access is necessary for determining how the interests and wishes of socioeconomically vulnerable people can be best respected as they make existential decisions about the ends of their lives. With the understanding that assisted dying has arrived on the global scene and is here to stay, this Note posits a comprehensive matrix of twelve categorical restrictions that will facilitate productive assessment of how lawmakers’ choices affect vulnerable people.

Footnotes
  1. ^ See Appendix, Harv. L. Rev., https://harvardlawreview.org/print/vol-139/barriers-and-safeguards-appendix, [https://perma.cc/YW22-3AL4].

    Return to citation ^
  2. ^ See Jennifer Clarke, How Could Assisted Dying Laws Change in England and Wales?, BBC (Mar. 18, 2026), https://www.bbc.com/news/articles/c5y5d2g3wgxo [https://perma.cc/LKH3-D5GA]; French Lawmakers Approve Assisted Dying Bill, Le Monde (May 27, 2025, at 20:56 CET), https://www.lemonde.fr/en/france/article/2025/05/27/french-lawmakers-approve-assisted-dying-bill_6741744_7.html [https://perma.cc/C79G-VV8N]; Uruguay’s Legislature Votes to Legalize Euthanasia, A First for South America, CBS News (Oct. 16, 2025, at 06:31 ET), https://www.cbsnews.com/news/uruguay-legalizes-euthanasia-first-country-south-america/ [https://perma.cc/BT94-QWK9].

    Return to citation ^
  3. ^ See Appendix, supra note 1.

    Return to citation ^
  4. ^ Coalition of Advocacy Organizations Across Illinois Hold Press Conference at Illinois State Capitol to Protest Legalization of Physician Assisted Suicide, Access Living (Mar. 13, 2024), https://www.accessliving.org/newsroom/press-releases-and-statements/coalition-of-advocacy-organizations-across-illinois-hold-press-conference-at-illinois-state-capitol-to-protest-legalization-of-physician-assisted-suicide/ [https://perma.cc/DJ83-HJFP] (quoting Vice President of Advocacy at Access Living Amber Smock). But see Ben Colburn, Disability‐Based Arguments Against Assisted Dying Laws, 36 Bioethics 680, 681 (2022) (“The evidence is . . . that people with disabilities, and disability rights organizations, have diverse views on the question of whether assisted dying should be legal.”).

    Return to citation ^
  5. ^ 11 Feb. 2025, Terminally Ill Adults (End of Life) Bill Deb (2025) cols. 361–64.

    Return to citation ^
  6. ^ Id. at cols. 363–64 (statement of Stephen Kinnock MP); id. at col. 361 (statement of Tom Gordon MP).

    Return to citation ^
  7. ^ 25 Feb. 2025, Terminally Ill Adults (End of Life) Bill Deb (2025) col. 492 (statement of Danny Kruger MP).

    Return to citation ^
  8. ^ See infra section II.A, pp. 1990–91.

    Return to citation ^
  9. ^ Agency for Healthcare Rsch. & Quality, National Healthcare Quality and Disparities Report, at A-11 (2021).

    Return to citation ^
  10. ^ Jane Rowley et al., The Impact of Poverty and Deprivation at the End of Life: A Critical Review, 15 Palliat. Care & Soc. Prac. 1, 8 (2021) (noting that “people living in the most deprived neighbourhoods [a]re less likely to receive specialist palliative care” at the end of life).

    Return to citation ^
  11. ^ Nat’l Acads. of Scis., Eng’g & Med., Health-Care Utilization as a Proxy in Disability Determination 23, 29–31 (2018).

    Return to citation ^
  12. ^ Cf. Anne Marie Su, Note, Physician Assisted Suicide: Debunking the Myths Surrounding the Elderly, Poor, and Disabled, 10 Hastings Race & Poverty L.J. 145, 167 (2013) (“If, on the other hand, the safeguards became overly stringent, such that no one could ever really qualify . . . , the law could be struck down later as an effective ban on [assisting dying].”).

    Return to citation ^
  13. ^ Andreas Fontalis, Efthymia Prousali & Kunal Kulkarni, Euthanasia and Assisted Dying: What Is the Current Position and What Are the Key Arguments Informing the Debate?, 111 J. Royal Soc’y Med. 407, 407–08 (2018). This is distinct from the practice of withdrawing life-sustaining treatment, which international law treats separately. See, e.g., Washington v. Glucksberg, 521 U.S. 702, 725 (1997) (distinguishing legal approaches to assisted dying from “the long legal tradition protecting the decision to refuse unwanted medical treatment,” and sanctioning only the latter).

    Return to citation ^
  14. ^ Sarah Mroz et al., Assisted Dying Around the World: A Status Quaestionis, 10 Annals Palliative Med. 3540, 3541 (2021).

    Return to citation ^
  15. ^ Appendix, supra note 1. But see Samia A. Hurst & Alex Mauron, Assisted Suicide and Euthanasia in Switzerland: Allowing a Role for Non-Physicians, 326 BMJ 271, 271 (2003).

    Return to citation ^
  16. ^ Fontalis, Prousali & Kulkarni, supra note 13, at 407–08.

    Return to citation ^
  17. ^ Meta Rus & Chris Gastmans, Euthanasia and Assisted Suicide: Who Are the Vulnerable?, 19 Clinical Ethics 18, 18 (2024); see Katie Morris, Vulnerability, Care Ethics and the Protection of Socioeconomic Rights via Article 3 ECHR, Hum. Rts. L. Rev., Dec. 2023, ngad028, at 1 (stating that, in the international human rights law context, vulnerability is “broadly construed as a state of being whereby one is open and exposed to hurts and harms of various kinds” (internal quotations omitted)).

    Return to citation ^
  18. ^ See, e.g., Alexandra Mullock & Jonathan Lewis, Assisted Dying, Vulnerability, and the Potential Value of Prospective Legal Authorization, Med. L. Rev., Spring 2025, fwaf014, at 4; Samia A. Hurst, Vulnerability in Research and Health Care; Describing the Elephant in the Room?, 22 Bioethics 191, 197 (2008).

    Return to citation ^
  19. ^ Martha Albertson Fineman, The Vulnerable Subject: Anchoring Equality in the Human Condition, 20 Yale J. L. & Feminism 1, 11–12 (2008); see also Mullock & Lewis, supra note 18, at 5.

    Return to citation ^
  20. ^ See, e.g., 11 Feb. 2025, Terminally Ill Adults (End of Life) Bill Deb (2025) col. 364 (statement of Stephen Kinnock MP).

    Return to citation ^
  21. ^ 521 U.S. 793 (1997).

    Return to citation ^
  22. ^ See, e.g., Brief of Amici Curiae States of California, Arkansas, Colorado, Florida, Georgia, Iowa, Maryland, Massachusetts, Michigan, Montana, Nebraska, South Carolina, Tennessee, Virginia, and Washington in Support of Petitioners Dennis C. Vacco, et al. at 14, Vacco v. Quill, 521 U.S. 793 (1997) (No. 95-1858).

    Return to citation ^
  23. ^ See David Moscrop, The Canadian State Is Euthanizing Its Poor and Disabled, Jacobin (May 2, 2024), https://jacobin.com/2024/05/canada-euthanasia-poor-disabled-health-care [https://perma.cc/ZVY7-ACX9].

    Return to citation ^
  24. ^ See, e.g., Maria Cheng & Angie Wang, Some Health Care Workers in Canada Grappling with Patients Requesting Euthanasia, PBS News (Oct. 16, 2024, at 12:59 ET), https://www.pbs.org/newshour/world/some-health-care-workers-in-canada-grappling-with-patients-requesting-euthanasia [https://perma.cc/9K5L-8LK8].

    Return to citation ^
  25. ^ 25 Mar. 2025, Terminally Ill Adults (End of Life) Bill Deb (2025) col. 1400.

    Return to citation ^
  26. ^ See Fontalis, Prousali & Kulkarni, supra note 13, at 410.

    Return to citation ^
  27. ^ Udo Schüklenk, Rethinking Assisted Dying, 41 Soc. Phil. & Pol’y 327, 329 (2024).

    Return to citation ^
  28. ^ Ellen R. Wiebe et al., Are Unmet Needs Driving Requests for Medical Assistance in Dying (MAiD)? A Qualitative Study of Canadian MAiD Providers, 47 Death Stud. 204, 205 (2023) (“Providing MAiD for someone whose unbearable suffering might be relieved by an expensive treatment they cannot access may cause moral distress.”); Katie Engelhart, Five Things to Know About Assisted Dying in Canada, N.Y. Times Mag. (June 1, 2025), https://www.nytimes.com/2025/06/01/magazine/medically-assisted-dying-canada-takeaways.html [https://perma.cc/P9LW-V7JL]; see Yukiko Asada et al., Importance of Investigating Vulnerabilities in Health and Social Service Provision Among Requestors of Medical Assistance in Dying, Lancet Reg’l Health - Ams., July 2024, at 1.

    Return to citation ^
  29. ^ Em Walsh, Parity, Poverty, and Physician Aid in Dying, Hastings Ctr. Rep., Sep.–Oct. 2024, at 24, 28.

    Return to citation ^
  30. ^ See Det Etiske Råds, Det Etiske Råds Udtalelse om Dødshjælp [The Danish Council of Ethics’ Statement on Euthanasia] 7 (2023) (stating that only an absolute ban would protect the lives and dignity of society’s most vulnerable); Fontalis, Prousali & Kulkarni, supra note 13, at 410 (“Rhetoric from opponents has raised concerns of whether such safeguards could ever be adequate, which appears to be a key argument in the debate . . . .”).

    Return to citation ^
  31. ^ James Downar, Susan MacDonald & Sandy Buchman, Medical Assistance in Dying, Palliative Care, Safety, and Structural Vulnerability, 26 J. Palliat. Med. 1175, 1176 (2023) (“In Canada, media widely reported the case of a woman with multiple chemical sensitivities . . . along with claims that she was driven to [assisted dying] through poverty and lack of adequate housing rather than intolerable suffering . . . .”).

    Return to citation ^
  32. ^ See Wiebe et al., supra note 28, at 204 (“Often, wealthy people with disabilities are able to access the equipment and care they need while those marginalized by poverty may not.”).

    Return to citation ^
  33. ^ See Lili Liu et al., Impacts of Poverty and Lifestyles on Mortality: A Cohort Study in Predominantly Low-Income Americans, 67 Am. J. Preventative Med. 15, 16 (2024) (noting that “[i]ndividuals with a low income are at an elevated risk of developing cardiovascular diseases, cancer and other noncommunicable diseases” and “also have worse disease prognoses” (footnotes omitted)).

    Return to citation ^
  34. ^ See Schüklenk, supra note 27, at 341.

    Return to citation ^
  35. ^ See id. at 340.

    Return to citation ^
  36. ^ See supra note 30 and accompanying text.

    Return to citation ^
  37. ^ Carter v. Canada (Att’y Gen.), 2012 BCSC 886, para. 1274 (Can. B.C.).

    Return to citation ^
  38. ^ [2015] 1 S.C.R. 331 (Can.).

    Return to citation ^
  39. ^ Schüklenk, supra note 27, at 343; see id. at 343–44 (citing and quoting Carter, 2012 BCSC 886, para. 1274).

    Return to citation ^
  40. ^ Mullock & Lewis, supra note 18, at 13.

    Return to citation ^
  41. ^ Id. at 13–14.

    Return to citation ^
  42. ^ Cf. Jasmine Cassy Mah et al., Social Vulnerability Indices: A Scoping Review, BMC Pub. Health, June 28, 2023, art. 1253, at 2 (describing aggregation as “[a] common way to estimate social vulnerability,” a measure similar to socioeconomic vulnerability).

    Return to citation ^
  43. ^ Dionysios Palermos et al., Individual-Based Socioeconomic Vulnerability and Deprivation Indices: A Scoping Review, Frontiers Pub. Health, Aug. 14, 2024, at 1. For example, the California Public Utilities Commission uses poverty rates, unemployment rates, educational attainment, proportion of income spent on housing, and linguistic isolation metrics to create its index. Cal. Pub. Utils. Comm’n, 2021/2022 Annual Affordability Report 11 (2023), https://www.cpuc.ca.gov/-/media/cpuc-website/divisions/energy-division/documents/affordability-proceeding/2021-2022/2021-and-2022-annual-affordability-report.pdf [https://perma.cc/5UGQ-DXK8]. The Canadian Index of Multiple Deprivation contemplates a wider range of factors, including the median value of local homes and proportion of local residents receiving government assistance. Stat. Can., The Canadian Index of Multiple Deprivation: User Guide, 2021 (Oct. 22, 2024), https://www150.statcan.gc.ca/n1/pub/45-20-0001/452000012023002-eng.htm [https://perma.cc/S9M4-JJFB].

    Return to citation ^
  44. ^ Appendix, supra note 1 (excepting Cuba, Estonia, Germany, Italy, Montana, and Switzerland).

    Return to citation ^
  45. ^ Compare End of Life Option Act, Cal. Health & Safety Code § 443.19(b) (West 2022) (requiring annual reporting on education levels and health insurance coverage, among other variables), with Loi du 28 mai 2002 relative à l’euthanasie [Law Relating to Euthanasia], M.B., June 22, 2002, art. 9, https://www.ejustice.just.fgov.be/cgi_loi/article.pl?language=fr&sum_date=&pd_search=2002-06-22&numac_search=2002009590 [https://perma.cc/R6FW-ZRUN] (requiring a biennial “description and evaluation of the application of [the] law,” without specified variables).

    Return to citation ^
  46. ^ See Sarah Scobie et al., Nuffield Tr., Assisted Dying in Practice app. (2025) (reporting various demographic data including education level, but no purely economic data).

    Return to citation ^
  47. ^ See supra notes 42–46 and accompanying text.

    Return to citation ^
  48. ^ For example, the California Socioeconomic Vulnerability Index uses unemployment, education, and other variables to supplement its description of socioeconomic status within specific census tracts. Cal. Pub. Utils. Comm’n, supra note 43, at 11.

    Return to citation ^
  49. ^ See Scobie et al., supra note 46, at app. (revealing that none of fifteen surveyed jurisdictions collect data on the incomes of assisted dying recipients). The research process underlying this Note found the same result as to all jurisdictions surveyed in the Appendix, supra note 1.

    Return to citation ^
  50. ^ Health Can., Fifth Annual Report on Medical Assistance in Dying in Canada 42–43, 43 fig. 4.5a (2024) [hereinafter Fifth Annual Report] (assessing all patients across Canada who received assisted dying in 2023); Health Can., Sixth Annual Report on Medical Assistance in Dying in Canada 42–43, 43 fig. 4.5a (2025) [hereinafter Sixth Annual Report] (same for 2024); Manny Tran et al., Socioeconomic Status and Medical Assistance in Dying: A Regional Descriptive Study, 37 J. Palliat. Care 359, 360, 362 (2022) (assessing patients in Ontario who lived within one particular hospital’s catchment area); Donald A. Redelmeier et al., Association of Socioeconomic Status with Medical Assistance in Dying: A Case-Control Analysis, BMJ Open, May 25, 2021, e043547, at 2–3 (assessing older adults who died while on a palliative care service in Ontario during a three-year period).

    Return to citation ^
  51. ^ Nicole Steck, Christoph Junker & Marcel Zwahlen, Increase in Assisted Suicide in Switzerland: Did the Socioeconomic Predictors Change? Results from the Swiss National Cohort, BMJ Open, Apr. 17, 2018, e020992, at 2–3, 6.

    Return to citation ^
  52. ^ Fifth Annual Report, supra note 50, at 42 (finding no conclusive evidence that any group of assisted dying recipients was disproportionately likely to occupy a relatively low income quintile); Sixth Annual Report, supra note 50, at 42–47 (same); Tran et al., supra note 50, at 362 (finding that patients within the lowest two income quintiles were more likely to request assisted dying, but not more likely to receive it); Redelmeier et al., supra note 50, at 3 (finding that patients with relatively low estimated incomes accessed assisted dying 39% less frequently than those with higher estimated incomes); Steck, Junker & Zwahlen, supra note 51, at 6 (finding that higher estimated income quintiles were associated with higher rates of access to assisted dying).

    Return to citation ^
  53. ^ Gerrit van der Wal et al., Medische besluitvorming aan het einde van het leven 69–70 (2003) (observing that individuals with “high” socioeconomic status, estimated via their residential postal codes, were more likely than other socioeconomic groups to receive assisted dying services, but restricting the analysis to those making end-of-life decisions, rather than all those who died); Jan H. Veldink et al., Euthanasia and Physician-Assisted Suicide Among Patients with Amyotrophic Lateral Sclerosis in the Netherlands, 346 NEJM 1638, 1641–43 (2002) (observing no statistically significant correlation between assisted dying and income or educational level, but restricting the analysis to a specific group of patients).

    Return to citation ^
  54. ^ Contrast Off. of the Chief Coroner, Ministry of the Solic. Gen., Medical Assistance in Dying (MAiD): Marginalization Data Perspectives 5, 6, 8 (2024) (noting that Ontarians in the most marginalized quintile of the “Material Resources dimension” were more likely to access assisted dying services, id. at 5, though this result became attenuated when comparing recipients only to people receiving palliative care services, id. at 8, and the difference could also be explained by correlations between advanced average age of recipients and certain marginalization indicators, like living alone, see id. at 5–6), with Fifth Annual Report, supra note 50, at 43–44 (finding that recipients across Canada are generally less likely to come from marginalized neighborhoods).

    Return to citation ^
  55. ^ Earnings by Educational Attainment, OECD, https://www.oecd.org/en/topics/sub-issues/earnings-by-educational-attainment.html [https://perma.cc/KLS4-SVHR] (annual income across nation-members of the Organisation for Economic Co-operation and Development); Education and Lifetime Earnings, Soc. Sec. Admin. (Nov. 2015), https://www.ssa.gov/policy/docs/research-summaries/education-earnings.html [https://perma.cc/FLJ6-8KD7] (U.S. lifetime earnings).

    Return to citation ^
  56. ^ See infra Table 1.

    Return to citation ^
  57. ^ See, e.g., Scobie et al., supra note 46, at app. (listing several jurisdictions that didn’t report educational data).

    Return to citation ^
  58. ^ See, e.g., Vict. State Gov’t, Voluntary Assisted Dying Review Board Annual Report: July 2024 to June 2025, at 20 (2025) (up to a third of entries did not include educational data).

    Return to citation ^
  59. ^ See, e.g., Margaret P. Battin et al., Legal Physician‐Assisted Dying in Oregon and the Netherlands: Evidence Concerning the Impact on Patients in “Vulnerable” Groups, 33 J. Med. Ethics 591, 594, 595 tbl. 1 (2007); Elissa Kozlov et al., Aggregating 23 Years of Data on Medical Aid in Dying in the United States, 70 J. Am. Geriatr. Soc’y 3040, 3042 tbl. 1 (2022).

    Return to citation ^
  60. ^ E.g., Medical Assistance in Dying (MAID), Stat. Can. (Nov. 28, 2025), https://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=5273 [https://perma.cc/S79P-QUB7] (describing the assisted dying dataset within the microdata access portal); Application Process and Guidelines – RDC, Stat. Can. (Mar. 20, 2026), https://www.statcan.gc.ca/en/microdata/data-centres/access [https://perma.cc/37AP-LJDV] (describing the process for requesting access to the secure microdata access portal).

    Return to citation ^
  61. ^ See, e.g., The Oregon Death with Dignity Act, Or. Rev. Stat. § 127.865(2) (2025); An Act to Enact the Maine Death with Dignity Act, Me. Stat. tit. 22, § 2140(17)(C) (2025). California law doesn’t specifically foreclose public access, but the state Department of Health clarified via email that it doesn’t disclose any data besides the summary statistics in its annual reports. Email from Cal. Dep’t of Pub. Health, to Student Author, Harv. L. Rev. (Aug. 4, 2025, at 16:42 ET) (on file with the Harvard Law School Library).

    Return to citation ^
  62. ^ End of Life Option Act, Cal. Health & Safety Code § 443.19(a) (West 2022).

    Return to citation ^
  63. ^ Table 1 is limited to individuals aged eighteen years and up because all U.S. legislative frameworks restrict assisted dying access to adults. See Appendix, supra note 1 (Montana doesn’t explicitly restrict access based on age, but the state has no legislative framework).

    Return to citation ^
  64. ^ See Appendix, supra note 1 (Montana, Oregon, and Vermont).

    Return to citation ^
  65. ^ E.g., Ctr. for Health Stat., Or. Health Auth., 2024 Oregon Death with Dignity Act Data Summary 13 tbl. 1 (2025) [hereinafter Oregon 2024] (reporting twenty-two (5.9%) nonresident cases in 2024, twenty-five (6.5%) in 2023, and three (0.1%) between 1998 and 2022).

    Return to citation ^
  66. ^ See, e.g., Cal. Dep’t of Pub. Health, California End of Life Option Act 2024 Data Report 6–8 (2025).

    Return to citation ^
  67. ^ See, e.g., End of Life Option Act, Cal. Health & Safety Code § 443.5(a) (West 2022) (describing eligibility review process).

    Return to citation ^
  68. ^ See, e.g., 2014–2024 Final Deaths by Year Statewide, CalHHS (Mar. 25, 2026), https://data.chhs.ca.gov/dataset/statewide-death-profiles/resource/f8bb67bf-923d-4d74-8714-dc2bcf5609b7 [https://perma.cc/K533-MLAZ].

    Return to citation ^
  69. ^ Employment Projections, U.S. Bureau Lab. Stat. (Aug. 28, 2025), https://www.bls.gov/emp/chart-unemployment-earnings-education.htm [https://perma.cc/L4RZ-JVD3].

    Return to citation ^
  70. ^ Additionally, several jurisdictions group high school diplomas and equivalencies together, see sources cited infra notes 76, 80, 83, 85, and Hawaii does not report less-than-high-school educational levels, see sources cited infra note 82.

    Return to citation ^
  71. ^ See John D. Parr et al., The Influence of Age on the Likelihood of Receiving End-of-Life Care Consistent with Patient Treatment Preferences, 13 J. Palliat. Med. 719, 719 (2010) (“As the US population ages a heightened understanding of how age influences health care for terminally ill patients is needed.”).

    Return to citation ^
  72. ^ Table 1 does not provide age 65+ reference values for the other two educational categories because the Census Bureau uses different categorical definitions across age groups. See 2023 American Community Survey 1-Year Estimates: Education, U.S. Census Bureau, https://data.census.gov/profile/United_States?g=010XX00US#education [https://perma.cc/U3X6-N5KZ].

    Return to citation ^
  73. ^ Contrast Kozlov et al., supra note 59, at 3042 (reporting median age of 74 years among assisted dying recipients in the U.S.), with 2023 American Community Survey 1-Year Estimates: Age and Sex, U.S. Census Bureau, https://data.census.gov/table/ACSST1Y2023.S0101?q=Age+and+Sex [https://perma.cc/2NDR-9CEA] (reporting median age of 39.2 years across the U.S.).

    Return to citation ^
  74. ^ U.S. Census Bureau, supra note 72. These data are sourced from the Census Bureau’s American Community Survey, which collects a range of demographic data via internet, mail, and in-person interviews from a stratified random sample of U.S. households. See U.S. Census Bureau, U.S. Dep’t of Com., American Community Survey and Puerto Rico Community Survey Design and Methodology 4-1, 4-14 (2024).

    Return to citation ^
  75. ^ Maine and Washington aren’t included because their reports don’t limit educational data reporting to individuals who died as a result of assisted dying services. See Me. Dep’t of Health & Hum. Servs. & Me. Ctr. for Disease Control & Prevention, The Maine Death with Dignity Act Statistical Report: 2024 Annual Report 4 (2025); Ctr. for Health Stat., Wa. State Dep’t of Health, 2023 Death with Dignity 5 tbl. 3 (2024).

    Return to citation ^
  76. ^ Cal. Dep’t of Pub. Health, California End of Life Option Act 2024 Data Report 13 tbl. 1 (2025); Cal. Dep’t of Pub. Health, California End of Life Option Act 2022 Data Report 10 tbl. 1 (2023).

    Return to citation ^
  77. ^ High School Diploma, GED, or Less, or Unreported. When a jurisdiction doesn’t include the relevant data within a particular annual report, this Note marks the entry in question as Unreported. When a jurisdiction hadn’t published a particular annual report as of the time this analysis was completed, this Note marks the entry in question as Unpublished.

    Return to citation ^
  78. ^ Some College or Associate’s Degree.

    Return to citation ^
  79. ^ Bachelor’s Degree or Higher.

    Return to citation ^
  80. ^ Colo. Ctr. for Health & Env’t Data, Colo. Dep’t of Pub. Health & Env’t, Colorado End-of-Life Options Act, 2024, at 6–7 tbl. 1 (2025).

    Return to citation ^
  81. ^ D.C. Health, District of Columbia Death with Dignity Act: 2022 Data Summary 2, 3 tbl. 1 (data from February 19, 2021 to February 18, 2022); D.C. Health, District of Columbia Death with Dignity Act: 2021 Data Summary 2, 3 tbl. 1 (data from February 19, 2020 to February 18, 2021).

    Return to citation ^
  82. ^ Off. of Plan., Pol’y, & Program Dev., Haw. Dep’t of Health, 2023 Our Care, Our Choice Act (OCOCA) Annual Report 5–8 (2024); Off. of Plan., Pol’y, & Program Dev., Haw. Dep’t of Health, 2022 Our Care Our Choice Annual Report 5–7 (2023); Off. of Plan., Pol’y, & Program Dev., Haw. Dep’t of Health, 2021 Our Care Our Choice Act Annual Report (DOH) 5–7 (2022).

    Return to citation ^
  83. ^ Off. of the Chief State Med. Exam’r, New Jersey Medical Aid in Dying for the Terminally Ill Act: 2024 Data Summary 6; Off. of the Chief State Med. Exam’r, New Jersey Medical Aid in Dying for the Terminally Ill Act: 2022 Data Summary 6; Off. of the Chief State Med. Exam’r, New Jersey Medical Aid in Dying for the Terminally Ill Act: 2021 Data Summary 6.

    Return to citation ^
  84. ^ Table 1 excludes this reporting period because officials stated that ninety-one patients completed its assisted dying process, but erroneously documented educational data from 101 patients (equal to the total number of cases filed). See Off. of the Chief State Med. Exam’r, New Jersey Medical Aid in Dying for the Terminally Ill Act: 2023 Data Summary 4, 6.

    Return to citation ^
  85. ^ Oregon 2024, supra note 65, at 12–13; Or. Health Auth., Oregon Death with Dignity Act: 2022 Data Summary 11 (2023).

    Return to citation ^
  86. ^ See Mark R. Phillips et al., The Clinician’s Guide to P Values, Confidence Intervals, and Magnitude of Effects, 36 Eye 341, 341 (2022).

    Return to citation ^
  87. ^ P(X33), where x=33 (number of 2022 assisted dying recipients in the “High School Diploma, GED, or Less, or Unreported” category); P=0.354 (proportion in the same category from the reference population), and n=91 (total number of 2022 assisted dying recipients).

    Return to citation ^
  88. ^ P(X122), where x=122, P=0.322, and n=376.

    Return to citation ^
  89. ^ Where P=0.352.

    Return to citation ^
  90. ^ P(X16), where x=16 and n=29.

    Return to citation ^
  91. ^ P(X18), where x=18 and n=37.

    Return to citation ^
  92. ^ P(X26), where x=26 and n=51.

    Return to citation ^
  93. ^ Where x=4 and n=17.

    Return to citation ^
  94. ^ Where x=8 and n=27.

    Return to citation ^
  95. ^ Where x=3 and n=28.

    Return to citation ^
  96. ^ Where x=7 and n=29.

    Return to citation ^
  97. ^ Where x=11 and n=37.

    Return to citation ^
  98. ^ Where x=5 and n=51.

    Return to citation ^
  99. ^ Sample sizes from D.C. were too small to support hypothesis testing.

    Return to citation ^
  100. ^ Based on imputation of missing data. See supra notes 96–98 and accompanying text.

    Return to citation ^
  101. ^ See supra notes 49–54 and accompanying text.

    Return to citation ^
  102. ^ As MP Atkinson feared. See supra note 25 and accompanying text.

    Return to citation ^
  103. ^ See Oregon 2024, supra note 65, at 13 (finding that in 2024, at least 83.1% of patients lived in or near urban centers and a further 5.9% lived out of state; in 2023, at least 82.7% lived in or near urban centers and a further 6.5% lived out of state).

    Return to citation ^
  104. ^ See supra notes 60–62 and accompanying text.

    Return to citation ^
  105. ^ Lindsey M. Freeman, Susannah L. Rose & Stuart J. Youngner, Poverty: Not a Justification for Banning Physician-Assisted Death, Hastings Ctr. Rep., Nov.–Dec. 2018, at 38, 42 (emphasis omitted).

    Return to citation ^
  106. ^ Id.

    Return to citation ^
  107. ^ Cf. id. at 43 (acknowledging that, in rare circumstances, family members may apply coercive pressure to patients making assisted dying decisions).

    Return to citation ^
  108. ^ See supra note 25 and accompanying text.

    Return to citation ^
  109. ^ Ute Eppinger, Commentary, Is Medically Assisted Suicide Feasible in Germany?, Medscape (Feb. 20, 2024), https://www.medscape.com/viewarticle/1000135 [https://perma.cc/BJ5Q-PH5M].

    Return to citation ^
  110. ^ Excepting Tuscany, which recently passed an implementing statute. Pier Francesco Bresciani, Assisted Suicide the Italian Way, Verfassungsblog (Mar. 31, 2025), https://verfassungsblog.de/assisted-suicide-italy [https://perma.cc/V2YA-9NP2].

    Return to citation ^
  111. ^ Katheryn Houghton, Getting a Prescription to Die Remains Tricky Even as Aid-in-Dying Bills Gain Momentum, KFF Health News (Mar. 30, 2021), https://kffhealthnews.org/news/article/montana-medical-aid-in-dying-legal-gray-zone-reviving-legislation [https://perma.cc/G8LJ-3SDX].

    Return to citation ^
  112. ^ Schweizerisches Strafgesetzbuch [StGB] [Criminal Code] Dec. 21, 1937, AS 54 757, as amended by Gesetz, June 23, 1989, AS 2449 (1989), art. 114–15.

    Return to citation ^
  113. ^ E.g., Criminal Code, R.S.C. 1985, c. C-46, § 241.2(1)(d), (3.1)(d).

    Return to citation ^
  114. ^ See generally Appendix, supra note 1 (citing statutes and regulations outlining these requirements).

    Return to citation ^
  115. ^ E.g., End of Life Option Act, Cal. Health & Safety Code § 443.5(a)(2) (West 2016).

    Return to citation ^
  116. ^ See Appendix, supra note 1.

    Return to citation ^
  117. ^ See id.

    Return to citation ^
  118. ^ See, e.g., Monique Kammeraat et al., Patients Requesting and Receiving Euthanasia for Psychiatric Disorders in the Netherlands, BMJ Mental Health, July 2023, at 1.

    Return to citation ^
  119. ^ See Appendix, supra note 1.

    Return to citation ^
  120. ^ E.g., Anita Hannig, Opinion, The Case for Assisted Dying, Newsweek (Feb. 13, 2024, at 15:08 ET), https://www.newsweek.com/case-assisted-dying-opinion-1869491 [https://perma.cc/J6CH-7YW2] (“The U.S. has the most restrictive assisted dying laws in the world.”).

    Return to citation ^
  121. ^ Harriet Sherwood, Dutch Woman, 29, Granted Euthanasia Approval on Grounds of Mental Suffering, The Guardian (May 16, 2024, at 21:30 ET), https://www.theguardian.com/society/article/2024/may/16/dutch-woman-euthanasia-approval-grounds-of-mental-suffering [https://perma.cc/6PA2-NBKN].

    Return to citation ^
  122. ^ HL Deb (12 Sep. 2025) (848) col. 1863 (statement of Lord Ashcombe MP).

    Return to citation ^
  123. ^ See Appendix, supra note 1 (providing a fulsome analysis of each jurisdiction’s extant framework and assessing for the presence or absence of each restriction).

    Return to citation ^
  124. ^ See generally Corte Constitucional [C.C.] [Constitutional Court], julio 22, 2021, Sentencia C-233/21 (extending eligibility to individuals without terminal prognoses).

    Return to citation ^
  125. ^ See First Colombian with Non-Terminal Illness Dies Legally by Euthanasia, Reuters (Jan. 8, 2022, at 19:42 ET), https://www.reuters.com/world/americas/first-colombian-with-non-terminal-illness-dies-legally-by-euthanasia-2022-01-08 [https://perma.cc/7AMU-TF6E].

    Return to citation ^
  126. ^ Kammeraat et al., supra note 118, at 4.

    Return to citation ^
  127. ^ One comparative international study revealed a significant correlation between criminalizing administration of assisted dying medication by someone other than the patient and lower uptake of assisted dying. See Brandon Heidinger et al., Research Letter, International Comparison of Underlying Disease Among Recipients of Medical Assistance in Dying, 185 JAMA Internal Med. 235, 236 (2025). Such measures exclude people who lose capacity after making a written request, see Christina Frangou, When Assisted Dying Means You Have to Go Before You’re Ready, The Guardian (Mar. 4, 2020, at 05:15 ET), https://www.theguardian.com/society/2020/mar/04/assisted-dying-maid-canada-leila-bell [https://perma.cc/N8BP-DN7S], and those physically unable to self-administer, Diane Marie et al., Navigating End-of-Life Decisions with Amyotrophic Lateral Sclerosis: A Patient-Centered Perspective on the Clinical and Legal Barriers to Medical Aid in Dying, Cureus, Sep. 13, 2025, at 1. Many patients prefer that the requisite medication be administered by someone else. See Scobie et al., supra note 46, at 48 (“[W]here self-administration is a requirement, there tends to be fewer assisted deaths, as a proportion of all deaths, than where clinician administration is an option . . . .”); Fifth Annual Report, supra note 50, at 14, 17 (reporting that, of 15,343 recipients in 2023, id. at 14, fewer than five chose to self-administer, id. at 17).

    Return to citation ^
  128. ^ Prognostic requirements correlate with reduced uptake of assisted dying. See Heidinger et al., supra note 127, at 236. Health care providers may be unable to accurately predict life expectancy, leading to arbitrary exclusions. See Scott A. Murray & Simon Noah Etkind, Editorial, Assisted Dying and the Difficulties of Predicting End of Life, BMJ, Mar. 12, 2025, r508, at 1. The scope of prognostic exclusion is also evidenced by marked utilization of the “Track 2” pathway, which the Canadian government introduced in 2021 to conditionally excuse the requirement that a patient’s death be “reasonably foreseeable” according to a qualified health care provider. An Act to Amend the Criminal Code (Medical Assistance in Dying), R.S.C. 2021, c. 2, cl. 3, 6. Since this amendment entered into force, 2,039 people have used Track 2. Sixth Annual Report, supra note 50, at 10.

    Return to citation ^
  129. ^ In jurisdictions that provide for mental illness eligibility, uptake is meaningful and consistent. In Belgium, 1.4% of assisted dying cases involve an underlying “psychiatric illness.” Press Release, Fed. Comm’n for the Control & Evaluation of Euthanasia, EUTHANASIA — Figures for 2023 (Feb. 27, 2024) [hereinafter Publication of the 2023 Figures], https://consultativebodies.health.belgium.be/en/documents/press-release-fccee-euthanasia-figures-2023 [https://perma.cc/75ZM-9YXT]; Press Release, Fed. Comm’n for the Control & Evaluation of Euthanasia, EUTHANASIA — Publication of the 2024 Figures for Euthanasia in Belgium (Mar. 19, 2025), https://consultativebodies.health.belgium.be/en/documents/euthanasia-publication-2024-figures-euthanasia-belgium [https://perma.cc/9VUQ-3XV9]. In the Netherlands, 1.5% and 2.5% of cases involved such diagnoses in 2023 and 2024, respectively. Reg’l Euthanasia Rev. Comms., Annual Report 2023, at 12, 14–15 (2024); Reg’l Euthanasia Rev. Comms., Annual Report 2024, at 14, 16–17 (2025).

    Return to citation ^
  130. ^ This measure excludes people who become unconscious, which is a common result of serious illness and/or palliative care, see Sophie Pautex et al., Letter, State of Consciousness During the Last Days of Life in Patients Receiving Palliative Care, 38 J. Pain & Symptom Mgmt. e1, e2 (2009) (finding that patients were unconscious for a median three days before death), and also those suffering from cognitive disorders like Alzheimer’s disease, see Paul T. Menzel & Bonnie Steinbock, Advance Directives, Dementia, and Physician-Assisted Death, 41 J.L. Med. & Ethics 484, 484 (2013). In Belgium and Canada, hundreds of people use advance directives to access assisted dying every year. See Publication of the 2023 Figures, supra note 129 (0.6% of recipients); Fifth Annual Report, supra note 50, at 14–17 (3.9% of recipients).

    Return to citation ^
  131. ^ While many diagnostic requirements may read expansively, they are a leading ground for determinations of ineligibility. In Canada, 57% of the reasons health care providers cite for rejecting assisted dying requests are diagnostic. See Fifth Annual Report, supra note 50, at 19–20. In New Zealand, 49.2% of reasons are diagnostic. See Health N.Z., Assisted Dying Service Annual Service Report 2024, at 10 (2025) [hereinafter New Zealand 2024].

    Return to citation ^
  132. ^ See Jozef H.H.M. Dorscheidt, The Legal Relevance of a Minor Patient’s Wish to Die: A Temporality-Related Exploration of End-of-Life Decisions in Pediatric Care, 45 Hist. & Phil. Life Scis., Jan. 17, 2023, at 7–8, 10–11 (reporting seventeen cases involving minors over a twenty-year period in the Netherlands and four cases over a five-year period in Belgium).

    Return to citation ^
  133. ^ See New Zealand 2024, supra note 131, at 10 (reporting that 3.6% of ineligibility decisions were based at least in part on citizenship or residency); Fifth Annual Report, supra note 50, at 19–20 (1.3%). In Switzerland, which does not impose this requirement, nearly a third of assisted dying recipients come from outside of the country. See Christine Bartsch et al., Assisted Suicide in Switzerland, 116 Deutsches Ärzteblatt Int’l 545, at supp. material IV tbl. 4 (2019).

    Return to citation ^
  134. ^ People living in rural settings may have to travel long distances to find two health care providers willing to confirm their eligibility. See Elise Bavazzano, Die Another Day: The Ineffectiveness of Medical Aid in Dying Legislation, 22 Rutgers J.L. & Religion 212, 222 (2021) (“[P]atients in rural areas have significantly less access than those in urban and suburban settings to find a doctor at all, let alone one willing to participate in [assisted dying].”); C. Ruth Wilson et al., Progress Made on Access to Rural Health Care in Canada, 66 Canadian Fam. Physician 31, 32 (2020) (noting that 18% of Canadians live in rural areas, but only 8% of physicians practice in rural areas).

    Return to citation ^
  135. ^ Committees reviewing hundreds or thousands of requests per year may experience delays, during which patients may die or lose capacity to consent. Cf. Kammeraat et al., supra note 118, at 4 (reporting a two-year average assessment time for patients seeking assisted dying on the basis of mental illness).

    Return to citation ^
  136. ^ Written declarations impose burdens on physically disabled individuals who cannot sign a declaration and must find an authorized representative to do so on their behalf.

    Return to citation ^
  137. ^ Patients may die or lose capacity during mandatory waiting periods. Even when waiting periods are waivable for good cause, see Appendix, supra note 1, they may inhibit access because patients can’t necessarily predict whether they’ll be entitled to a waiver. See Frangou, supra note 127 (profiling a patient with Alzheimer’s who ended her life earlier than she wished).

    Return to citation ^
  138. ^ These restrictions can diminish patients’ perceived privacy as to deeply personal choices about how their lives will end. Cf. Johnna P. Wellesley, Legal Medical Aid in Dying: The Paradox of Privacy, Hastings Ctr. for Bioethics (Oct. 27, 2025), https://www.thehastingscenter.org/legal-medical-aid-in-dying-the-paradox-of-privacy [https://perma.cc/X8S8-S5MX] (describing government records of assisted deaths from “socially problematic or value-laden condition[s] or method[s]” as “final, bureaucratic mark[s] of stigma”).

    Return to citation ^
  139. ^ Engelhart, supra note 28; accord Nadine Yousif, Assisted Dying Now Accounts for One in 20 Canada Deaths, BBC (Dec. 12, 2024), https://www.bbc.com/news/articles/c0j1z14p57po [https://perma.cc/LH54-Q96P].

    Return to citation ^
  140. ^ Editorial, Canada, There’s Still Time to Rethink this Risky Expansion of Euthanasia,Wash. Post (Jan. 27, 2024), https://www.washingtonpost.com/opinions/2024/01/27/canada-maid-euthanasia-death-assisted-suicide [https://perma.cc/KH6E-ANN5].

    Return to citation ^
  141. ^ See Appendix, supra note 1.

    Return to citation ^
  142. ^ See id. (Colombia, Estonia, Germany, Italy excepting Tuscany, and Montana).

    Return to citation ^
  143. ^ The Right to Die in Belgium: An Inside Look at the World’s Most Liberal Euthanasia Law, PBS News (Jan. 15, 2015, at 17:53 ET), https://www.pbs.org/newshour/show/right-die-belgium-inside-worlds-liberal-euthanasia-laws [https://perma.cc/F372-ZYDS]; see also Ashley Frawley, Euthanasia Isn’t Only a Matter of Personal Autonomy, Brussels Times (Mar. 26, 2025), https://www.brusselstimes.com/1505532/euthanasia-isnt-only-a-matter-of-personal-autonomy [https://perma.cc/KA3U-S9V6] (“Canada is often perceived as an outlier, but Belgium’s laws are actually more permissive . . . .”).

    Return to citation ^
  144. ^ Excluding Canada, where the federal Parliament has twice delayed the introduction of mental illness as a ground for access. Appendix, supra note 1.

    Return to citation ^
  145. ^ Id.

    Return to citation ^
  146. ^ Engelhart, supra note 28 (citing An Act to Amend the Criminal Code (Medical Assistance in Dying), R.S.C. 2021, c. 2, cl. 3).

    Return to citation ^
  147. ^ Kalima Carrigan, One-Way Ticket to Zürich: Presentations of “Suicide Tourism” in European News Media, 29 Mortality 817, 828 (2024).

    Return to citation ^
  148. ^ See supra sections II.A, pp. 1990–91, and II.B, pp. 1992–99.

    Return to citation ^