Contextualizing attitudes toward medical aid in dying in a national sample of interdisciplinary US hospice clinicians: hospice philosophy of care, patient-centered care, and professional exposure

Abstract

Background:

Despite extensive theoretical debate, empirical research on medical aid in dying (MAID) largely has disregarded broader, contextual factors as potential correlates of attitudes in hospice clinicians.

Objective:

Informed by institutional theory and neofunctional attitude theory, the objective of the current study was to quantitatively examine hospice clinicians’ attitudes toward MAID as functions of institutional characteristics relating to (Aim 1) individual adherence to hospice values and (Aim 2) state law.

Design:

We used a cross-sectional design.

Methods:

A national convenience sample of interdisciplinary hospice clinicians recruited through US professional membership associations self-administered an online survey. Measures included attitudes toward MAID, attitudes toward the hospice philosophy of care, attitudes toward the principle that hospice care should not hasten death, orientation toward patient-centeredness, professional exposure to working in a state where MAID is legal, and demographic characteristics. Data were analyzed via a partial proportional odds model.

Results:

The sample (N = 450) comprised hospice physicians (227 [50.4%]), nurses (64 [14.2%]), social workers (74 [16.4%]), and 85 chaplains (85 [18.9%]). Results of the partial proportional odds model indicated that professional exposure to working in a state where MAID is legal was significantly associated with over twice the cumulative odds of MAID support. Although neither orientation toward patient-centered care nor attitudes toward the hospice philosophy of care was significantly associated with attitudes toward MAID, results showed that disagreement with the narrower principle that hospice care should not hasten death was significantly associated with 6-to-7 times the cumulative odds of MAID support.

Conclusion:

Findings suggest that contextual factors—namely, the environments in which hospice clinicians practice—may shape attitudes toward MAID. Unanticipated results indicating that hospice professionals’ adherence to hospice values was not significantly associated with attitudes toward MAID underscore the need for further research on these complex associations, given previous theoretical and empirical support.

Introduction

Medical aid in dying (MAID) presents clinical, ethical, and legal implications for end-of-life care. In the United States, MAID describes a qualified person’s self-administration of a medication prescribed by an attending provider specifically to hasten their death. MAID is currently legal via statute in 9 states and Washington, DC. ¹ These jurisdictions account for >72 million individuals corresponding to one-fifth of the country’s population, ² with further legalization projected. ¹

MAID’s pertinence to the end of life has situated hospice care at the center of ongoing discussion. US policy outlines hospice care as a program for the provision of a specific type of palliative care for individuals with prognoses of ⩾6 months.^3,4 Hospice care’s recognition as gold-standard end-of-life care has contributed to its characterization as a societal mechanism fostering responsible US MAID implementation^5,6: Not only do MAID statutes require providers to educate patients on hospice care’s ability to palliate remediable symptoms, but data have shown that most (87%) of those who use MAID are hospice patients. ⁷

Despite this overlap, knowledge about hospice clinicians’ attitudes toward MAID is lacking. Research on correlates of hospice clinicians’ attitudes has prioritized individual-level characteristics. Personal characteristics linked to higher support include younger age,^{8 –10} gender identity as a man,^{8,10 –12} lack of religious affiliation,^12,13 and lower religiosity.^{8,10 –16} Exploration of professional characteristics has found increased years working in hospice or palliative care to be associated with decreased MAID support. ¹¹ Additionally, the limited research comparing professional disciplines ¹⁷ has indicated that, although physicians are more supportive of MAID than nurses,^{18 –20} social workers are more supportive than physicians^18,19,21 and nurses.^18,21

Conceptualizing attitudes as solely individual derivatives neglects institutional characteristics that may unify these interdisciplinary healthcare professionals as hospice clinicians. Doing so hinders understanding of how MAID attitudes fit within the context of US hospice care. This obscurity is noteworthy because hospice clinicians are embedded within a distinct approach to care that advances core institutional values, namely the hospice philosophy of care and patient-centered care. ²² At the same time, external policies may constrain clinicians’ ability to provide care aligned with their interpretations of such values. Despite this conceptual relevance, associations between (1) the hospice philosophy of care, patient-centered care, and state law, and (2) attitudes toward MAID appear yet to be quantitatively tested.

Theoretical background

Institutional theory

Institutional theory and neofunctional attitude theory provide a frame for conceptualizing how contextual factors shaping the hospice institution may guide clinicians’ approach to the provision of hospice care. Institutional theory elucidates how institutions and their environments shape organizational functioning. ²³ Institutions represent social structures whose persistence is societally legitimized by regulative, normative, and cultural–cognitive pillars. ²³ Institutions engender institutional logics, or socially constructed operating principles that provide meaning to organizations’ and their employees’ daily activities, such as beliefs, rules, and values. ²⁴ As physical entities, organizations provide infrastructure to advance institutions. ²³ This infrastructure includes individual actors, who script institutional logic into organizational structure and behavior. ²³ Increased adherence to institutional scripts lends organizations increased societal legitimacy, which fosters their continued survival. ²³

Organizations also exist in perpetual and reciprocal interaction with the institutional environments in which they are situated. This interaction means that institutions simultaneously reflect and reproduce existing institutional fields. Social movements advocating for policy change, such as the death with dignity movement, exemplify such processes by reflecting societal renegotiations of obsolescent institutional logics. ²⁵ Thus, institutionalization of new policy produces a reconfigured institution, ²⁵ which itself confers wider sociopolitical legitimacy and obliges organizations and actors to operate within novel institutional environments. ²³

Neofunctional attitude theory

Neofunctional attitude theory proposes that attitudes, or evaluations of a given object, serve instrumental purposes by conferring psychological benefits to holders. ²⁶ The benefit derived varies by the specific function that the attitude serves. ²⁶ The value-expressive attitude function postulates that attitudes affirm individuals’ self-concept by representing important values.^26,27 Values are broadly applicable, abstract ideals that inform specific evaluations. ²⁷ Not only has the implication of values in virtually all social problems ²⁸ led to their general recognition as chief determinants of attitudes, ²⁹ but theorists have asserted that value-expressive attitudes should correspond more strongly to values than attitudes serving other attitude functions. ²⁷ The contrast between the individual-level focus of attitudes ²⁹ and the prospect of shared values ²⁷ is especially topical to organizational contexts, where actors are assumed to espouse the institutional values advanced through their organization.

Conceptual application

These two theories trace an ecological throughline that brings hospice clinicians’ MAID attitudes into sharper focus. Internally, institutional theory ²³ suggests that hospice organizations and clinicians promote values-based institutional logics corresponding to the hospice philosophy of care and patient-centered care. Neofunctional attitude theory’s ²⁶ value-expressive attitude function further suggests that hospice clinicians’ attitudes toward MAID are expressions of their espousal and interpretation of these values. Externally, however, institutional theory ²³ suggests that relevant policies influence hospice clinicians’ attitudes toward MAID by conditioning the institutional environment in which MAID attitudes are developed and maintained.

Empirical background

Hospice values

Hospice philosophy of care

The hospice philosophy of care is a set of values guiding the provision of quality hospice care. ³⁰ The National Hospice and Palliative Care Organization ²² recognizes Kirk’s ³⁰ four-point distillation of Saunders’s ³¹ original 13 principles of hospice care as constituting the modern hospice philosophy of care. These elements include dying as an experience, family-centered care, the nature and relief of suffering, and restoring and supporting moral agency. ³⁰

Absent from this recent iteration is the principle that hospice care neither prolongs nor hastens death. ³² This omission suggests an evolution that directly implicates MAID. Scholars have contended that MAID has contributed to an international^33,34 “mission creep,” ³⁵ which has led the hospice philosophy of care away from its original ethos. Given that the hospice philosophy of care has driven hospice and palliative care institutional position statements,^{34,36 –38} hospice organizational policies,^{39 –41} and hospice clinicians’ attitudes,^42,43 conceptualizations thereof have been noted to assume an important role in the downstream provision of care. ³⁴

Despite extensive theoretical debate on the hospice philosophy of care’s congruence with MAID, limited research has considered this association empirically. Research has illuminated a history of divergent attitudes toward MAID based on varying interpretations of the hospice philosophy of care,^{34,36,41,44 –46} even predating the first US legalization. ⁴⁷ Some studies have found contested interpretations of the same principles, such as the nature and relief of suffering, to undergird the stances of both opposition and support.^39,41,48 Others have found tensions among principles. For instance, objections to MAID integration in hospice care premised on the former principle that hospice care does not hasten death^{36,41,44 –46} may contradict support based another principle, the relief of suffering.^{49 –51} Dissensus about the scope and rank of individual principles have generated conclusions that the principles grounding the hospice philosophy of care “are not in the context of [MAID] internally consistent” ⁴⁸ resulting in uncertainty regarding how the hospice philosophy of care corresponds to clinicians’ attitudes toward MAID.

Patient-centered care

Patient-centered care is defined as “care that is respectful of and responsive to individual patient preferences, needs, and values and ensuring that patient values guide all clinical decisions.” ⁵² Patient-centered care does not connote health care professionals’ abdication of medical decision-making to patients; rather, it brings patient preferences to the forefront of a collaborative endeavor. ⁵³

Appropriate foregrounding of patient preferences in end-of-life treatment planning has formed the basis of patient-centered care-based support for MAID.^54,55 In addition to explicit references to patient-centeredness in clinicians’ support for MAID,^42,46,50 qualitative,^46,50,56,57 and quantitative^14,58 studies have demonstrated that tenets of patient-centered care (viz, respect for a patient’s autonomy, recognition of the patient’s right to guide appropriate clinical decision-making) were positively related to MAID support. Nevertheless, direct associations between patient-centered care and MAID attitudes have not been assessed.

State law

MAID research often disregards differences in external environments by presenting participants as one monolithic sample. Perhaps the most substantial difference implicated by external environments is MAID legalization. The United States’ state-based approach to legalization exposes clinicians to practice environments permitting drastically different treatment options, which, thus, may inform attitude formation and change. ⁴²

Professional exposure

Previous research has largely stopped short of conceptualizing practice environments in relation to MAID’s legal standing.^{10,16,58 –61} It appears that only one study has examined the association between state policy and MAID attitudes. ⁶² This study found that nurses across specialties who lived in states where MAID was legal were more supportive of MAID than those who did not. ⁶² Nevertheless, this result does not account for where patients receive hospice care, which may be pertinent to hospices with service areas traversing state borders where MAID legalization differs. Additionally, it does not speak to whether or not this association holds across the hospice interdisciplinary group.

Study objective, aims, and hypotheses

Responding to calls for increased consideration of contextual correlates of clinicians’ MAID attitudes,^34,63 the objective of the current study was to examine hospice clinicians’ attitudes toward MAID as functions of institutional characteristics relating to (Aim 1) individual adherence to hospice values and (Aim 2) state law. Informed by institutional theory’s concept of institutional logics and neofunctional attitude theory’s value-expressive attitude function, Aim 1’s hypotheses stated:

• H ₁  Attitudes toward the hospice philosophy of care will be significantly associated with attitudes toward MAID.

• H ₂  A stronger orientation toward patient-centeredness will be significantly associated with more supportive attitudes toward MAID.

The use of the hospice philosophy of care to bolster positions of support and opposition^{34,36,39,41,44 –46,48} precluded formulating a directional hypothesis. Drawing on institutional theory’s notion of institutional environments, Aim 2’s hypothesis stated:

• H ₃  Participants with professional exposure to working in states where MAID is legal will demonstrate significantly more supportive attitudes toward MAID than those with professional exposure to working only in states where MAID is illegal.

Methods

Study design

This cross-sectional study entailed recruiting interdisciplinary US hospice clinicians to self-administer a one-time Qualtrics survey. Inclusion criteria stipulated that participants were ⩾18 years old, worked as paid hospice employees, and provided direct patient care. We employed a gatekeeper-facilitated convenience sampling approach targeting the listed members of national hospice and palliative care professional membership associations representing each of the core disciplines of the hospice interdisciplinary group (unnamed as per research agreement stipulations). ³ Internal recordkeeping for the medicine, nursing, and spiritual care professional membership associations precluded excluding nonhospice palliative care members from the sampling frame.

Research agreements executed with participating professional membership associations varied in terms of permitted data collection procedures. Thus, we tailored each professional membership association’s survey dissemination to maximize response. Doing so prompted variation in the number of survey invitations (2 vs 5), dissemination mode (email vs newsletter), and dissemination source (professional membership organization vs research team). Despite implementing all appropriate security features, ⁶⁴ receipt of an anomalously high number of responses (n = 2392) immediately following the second invitation in the nursing survey raised suspicion of bot infiltration. Therefore, we listwise deleted any cases in the nursing survey with a start date occurring on or after the second invitation. Our compensatory round of data collection generated no concerns about fraud.

Data collection occurred from November 2022 through January 2023. Surveys remained open for 30 days. We compensated 200 randomly selected participants via $20 egift cards. We defined survey break-off through established cutoffs for the percentage of all unanswered survey items (⩽49%). ⁶⁵ Our reporting of quality criteria herein conforms to Strengthening the Reporting of Observational Studies in Epidemiology ⁶⁶ guidelines for cross-sectional studies (see Supplemental Table 1).

Measures

Dependent variable

Attitudes toward MAID

We measured attitudes toward MAID through a researcher-constructed Likert-type item (1 = strongly oppose to 5 = strongly support) asking, “To what extent do you support or oppose medical aid in dying for a patient with an expected 6 months or fewer to live?” Item wording, ⁶⁷ response options,^68,69 and conceptual definitions provided to distinguish MAID from euthanasia were informed by previous research.^70,71 Conceptual definitions were presented as follows:

• MAID: The supply or prescription of drugs by a medical provider to be taken by the patient, with the explicit intention of hastening the patient’s death, at that competent patient’s voluntary and explicit request.

• Euthanasia: The administration of drugs by a clinician to the patient with the explicit intention of hastening the patient’s death, at that competent patient’s voluntary and explicit request.

Independent variables

Attitudes toward the hospice philosophy of care

We measured attitudes toward hospice philosophy of care using the eight-item Hospice Philosophy Scale (HPS-8). ⁷² HPS-8 items are scored on a Likert scale from 1 (strongly disagree) to 5 (strongly agree). Higher sum scores (theoretical range, 8–40) indicate higher support for the hospice philosophy of care. As with previous studies,^72,73 we randomly sequenced item presentation. The HPS-8 has demonstrated acceptable-to-excellent internal consistency reliability,^{72 –75} convergent validity, ⁷² and the lack of substantive differential item functioning by age, gender, race, and hospice profession. ⁷⁴ Internal consistency reliability in the current study (α = .80) was good. ⁷⁶

Attitudes toward the principle that hospice care should not hasten death

We measured attitudes toward the principle that hospice care should not hasten death through a researcher-constructed item. This item read: “Please rate your agreement or disagreement with the following statement: Hospice care should not hasten death.” Data sparseness prompted collapsing the endpoints of the original 5-point Likert-type response scale (0 = agree to 2 = disagree).

Orientation toward patient-centeredness

We measured orientation toward patient-centeredness through the Patient–Practitioner Orientation Scale (PPOS), ⁷⁷ following permission from the scale developer (E. Krupat, email, 21 February 2021). The PPOS contains 18 items distributed equally across two subscales (Sharing, Caring). Items are scored on a Likert scale from 6 (strongly disagree) to 1 (strongly agree). Three items are reverse coded so that higher PPOS mean scores (theoretical range, 1–6) indicate a stronger orientation toward patient-centeredness. The PPOS has shown internal consistency reliability,^75,77 15-day test–retest reliability, ⁷⁸ construct validity,^{77 –80} content validity, ⁷⁸ and face validity.^78,79 We followed previous research ⁸¹ by slightly modifying item wording to improve hospice applicability. Internal consistency reliability in the current study (α = .61) was questionable. ⁷⁶

Professional exposure to working in a state where MAID is legal

We measured professional exposure to working in a state where MAID is legal through a researcher-constructed proxy item asking participants to select any states within their specific office’s service area. We triangulated responses with the Death with Dignity National Center’s ⁸² website to determine each state’s contemporaneous MAID legalization status. We coded responses to infer whether or not participants had any professional exposure to working in a state where MAID was legal at the time of data collection.

Covariates

Covariates included age in years, gender (woman, man), religious identity (Protestant, Catholic, other Christian, unaffiliated, other), religiosity (see below), years working in hospice, and profession (physician, chaplain, nurse, social worker). Excluding religiosity, all covariates were researcher-constructed single-item measures.

Religiosity

We measured religiosity through the brief version of the Santa Clara Strength of Religious Faith Questionnaire. ⁸³ The brief version’s five items are scored on a Likert scale from 1 (strongly disagree) to 4 (strongly agree). Higher sum scores (theoretical range, 5–20) indicate higher religiosity. The brief version has exhibited excellent internal consistency reliability,^84,85 2-week test–retest reliability, ⁸⁴ and convergent validity.^84,85 Internal consistency reliability in the current study (α = .94) was excellent. ⁷⁶

Statistical analysis

All analyses were conducted in Stata (Version 18; StataCorp: College Station, TX) with statistical significance defined as 2-tailed results of P < .05. First, we employed univariable statistics to characterize the sample and primary variables. Next, we performed chi-square tests to examine the bivariable associations between each categorical independent variable and MAID attitudes. Then, we examined bivariable differences in continuous independent variables by levels of MAID attitudes via either 1-way analyses of variance or Kruskal-Wallis tests adjusted for ties. These analyses were complemented by post hoc pairwise comparisons. The former used Bonferroni tests. The latter used Dunn tests with Benjamini-Hochberg adjustments to control for false discovery rates.

Last, we examined multivariable associations with MAID attitudes through a partial proportional odds model, using the user-written gologit2 command. ⁸⁶ Proportional odds models examine the associations of independent variables across thresholds of an ordinal dependent variable.^86,87 They compare the baseline cumulative odds of being at or below a given level versus the cumulative odds of being above that level for each threshold.^86,87 Observation of nonconvergence and negative predicted probabilities resulted in collapsing dependent variable endpoints. ⁸⁷ Thus, the final model imposed two thresholds: (Threshold 1) at or below oppose versus above oppose and (Threshold 2) at or below neither support nor oppose versus above neither support nor oppose. Assumption checks indicated that slopes for attitudes toward the hospice philosophy of care and the nursing profession varied across thresholds of attitudes toward MAID. Therefore, we freed these variables from the proportional odds constraint.^86,87

Results

Sample specification

Individual gatekeeper-provided membership estimates (range, 853–8500) indicated a sampling pool of 20 075 individuals. We counted 1346 survey responses. We specified our sample via a three-step process. First, we excluded cases based on prespecified data quality measures. Specifically, we screened 130 cases for lacking informed consent, 683 cases for not meeting inclusion criteria, 26 cases for constituting survey break-off, and 11 cases for not meeting Qualtrics’s security measures.

Second, we excluded cases based on conceptual appropriateness for sample retention, following a manual review of the data. We screened 7 cases to raise concerns about fitness in our theoretical population (eg, non-US clinicians, pediatric clinicians). We subsequently screened 6 cases for suspected misinterpretation of the select-all-that-apply proxy item assessing professional exposure to working in a state where MAID is legal. Responses seemed to reflect states where participants’ hospice organization has offices instead of states within participants’ employing hospice office’s service area. These cases raised suspicion by listing states separated by geographic distances rendering individual travel unrealistic (eg, California and Florida). We then screened 8 cases with sparse data in certain variable categories when conceptual differences precluded collapsing (ie, gender outside of “man” or “woman,” professional discipline outside of “chaplain,” “nurse,” “physician,” or “social worker”).

Third, we excluded cases based on missingness. The amount of the resulting 475 cases containing any missing data on variables in the analysis model was low (5%). Missingness ranged from a minimum of less than 1% on select items in the HPS-8, PPOS, and brief version of the Santa Clara Strength of Religious Faith Questionnaire to a maximum of 3% on religious identity. We handled missingness through listwise deleting these 25 cases after the results of Little’s test indicated data to be missing completely at random ( ${\chi }_{205}^2$  = 217.42, P = .263).

Univariable

Sample characteristics

The sample comprised 450 hospice clinicians, half of whom were physicians (227 [50.4%]) and the other half of whom were split among chaplains (85 [18.9%]), social workers (74 [16.4%]), and nurses (64 [14.2%]; see Table 1). Age ranged from 24 to 78 years (median [interquartile range], 52.0 [42.0–62.0] years). Approximately two-thirds of participants were women (294 [65.3%]). The majority of participants identified as not Hispanic or Latino (428 of 445 [96.2%]) and White (391 of 447 [87.5%]). Just over one-quarter of the sample (122 [27.1%]) identified as Protestant. Participants indicated having worked in hospice care between 0 and 45 years (median [interquartile range], 8.0 [4.0–15.0] years). Employing hospices were largely nonprofit (321 of 449 [71.5%]) and religiously unaffiliated (370 of 449 [82.4%]), with most containing a policy outlining permissible employee MAID participation (235 [52.2%]).

Table 1.

Sample characteristics (N = 450).

Characteristic	No. (%)
Participant
Age, mean (SD), years	51.8 (12.4)
Gender
Man	156 (34.7)
Woman	294 (65.3)
Ethnicity a
Hispanic or Latino	17 (3.8)
Not Hispanic or Latino	428 (96.2)
Raceb,c
Asian	22 (4.9)
Black or African American	19 (4.3)
White	391 (87.5)
Multiracial	12 (2.7)
Other	3 (0.7)
Religious identity
Catholic	67 (14.9)
Protestant	122 (27.1)
Other Christian	91 (20.2)
Unaffiliated	102 (22.7)
Other	68 (15.1)
Years working in hospice, mean (SD)	10.7 (8.9)
Profession c
Chaplain	85 (18.9)
Nurse	64 (14.2)
Physician	227 (50.4)
Social worker	74 (16.4)
Hospice
Religious affiliation d
Yes	79 (17.6)
No	370 (82.4)
Tax status d
For profit	110 (24.5)
Government	18 (4.0)
Nonprofit	321 (71.5)
Policy outlining permissible employee medical aid in dying participation c
Yes	235 (52.2)
No	100 (22.2)
Unsure	87 (19.3)
Other	28 (6.2)

^an = 445.

^bn = 447.

^cCategory percentages do not sum to 100%, due to rounding.

^dn = 449.

Primary variables

Over half of the sample supported MAID (250 [55.6%]; oppose: 139 [30.9%]; neither support nor oppose: 61 [13.6%]). In addition to high support for the hospice philosophy of care (mean [SD], 36.3 [3.7]; median [interquartile range], 37.0 [35.0–39.0]), more than half of participants agreed with the principle that hospice care should not hasten death (269 [59.8%]; neither agree nor disagree: 139 [30.9%]; disagree: 42 [9.3%]). The sample reported a strong orientation toward patient-centeredness (mean [SD], 5.0 [0.4]; median [interquartile range], 5.0 [4.8–5.3]). Just over one-quarter of the sample reported professional exposure to working in a state where MAID is legal (120 [26.7%]).

Bivariable

Results from a series of chi-square tests indicated that MAID attitudes were significantly associated with attitudes toward the principle that hospice care should not hasten death ( ${\chi }_4^2$  = 86.66, P < .001), professional exposure to working in a state where MAID is legal ( ${\chi }_2^2$  = 18.04, P < .001), gender ( ${\chi }_2^2$  = 48.09, P < .001), religious identity ( ${\chi }_8^2$  = 68.38, P < .001), and profession ( ${\chi }_6^2$  = 41.09, P < .001; see Table 2). Specifically, higher percentages of support for MAID were seen in participants who disagreed (81.0%) or who neither disagreed nor agreed that hospice should not hasten death (81.3%), had exposure to working in states where MAID was legal (71.7%), were women (67.0%), were religiously unaffiliated (87.3%), and were social workers (81.1%), respectively.

Table 2.

Chi-square test associations with attitudes toward medical aid in dying (N = 450).

Variable	Oppose (n = 139)	Neither support nor oppose (n = 61)	Support (n = 250)	χ2 (df)	P value
	No. (%)	No. (%)	No. (%)
Hospice care should not hasten death				86.66 (4)	<.001
Agree	123 (45.7)	43 (16.0)	103 (38.3)
Neither agree nor disagree	13 (9.4)	13 (9.4)	113 (81.3)
Disagree	3 (7.1)	5 (11.9)	34 (81.0)
Professional exposure to working in a state where medical aid in dying is legal				18.04 (2)	<.001
Yes	21 (17.5)	13 (10.8)	86 (71.7)
No	118 (35.8)	48 (14.6)	164 (49.7)
Gender				48.09 (2)	<.001
Man	77 (49.4)	26 (16.7)	53 (34.0)
Woman	62 (21.1)	35 (11.9)	197 (67.0)
Religious identity				68.38 (8)	<.001
Catholic	32 (47.8)	9 (13.4)	26 (38.8)
Protestant	48 (39.3)	19 (15.6)	55 (45.1)
Other Christian	37 (40.7)	12 (13.2)	42 (46.2)
Unaffiliated	8 (7.8)	5 (4.9)	89 (87.3)
Other	14 (20.6)	16 (23.5)	38 (55.9)
Profession				41.09 (6)	<.001
Chaplain	31 (36.5)	13 (15.3)	41 (48.2)
Nurse	14 (21.9)	3 (4.7)	47 (73.4)
Physician	85 (37.4)	40 (17.6)	102 (44.9)
Social worker	9 (12.2)	5 (6.8)	60 (81.1)

One-way analyses of variance results indicated that orientation toward patient-centeredness did not significantly differ by levels of MAID attitudes (F_2,449 = 1.80, P = .166; see Table 3). However, a statistically significant difference in age was observed (F_2,449 = 9.02, P < .001). Bonferroni post hoc tests revealed that participants who supported MAID were, on average, 5.5 years younger than those who opposed MAID (mean difference = −5.5, P < .001).

Table 3.

Kruskal-Wallis test differences by levels of attitudes toward medical aid in dying (N = 450).

Variable	Oppose (n = 139)	Neither support nor oppose (n = 61)	Support (n = 250)	χ2 (df)	PBH value
	No. (rank sum)	No. (rank sum)	No. (rank sum)
Attitudes toward the hospice philosophy of care	139 (35,227.0)	61 (11,677.0)	250 (54,571.0)	11.54 (2)	.003
Orientation toward patient-centeredness a	0.4 (5.0)	0.5 (5.0)	0.4 (5.0)	1.80 (2, 449)	.166
Age a	11.7 (55.4)	13.1 (51.8)	12.3 (49.9)	9.02 (2, 449)	<.001
Religiosity	139 (43,339.0)	61 (13,865.0)	250 (44,271.0)	96.51 (2)	<.001
Years working in hospice	139 (37,552.5)	61 (12,801.0)	250 (51,121.5)	23.88 (2)	<.001

^aStatistics reflect a 1-way analysis of variance and are presented as mean, SD, mean, SD, mean, SD, F, df, and P, respectively.

P_BH value, Benjamini-Hochberg-adjusted statistical significance of the Kruskal-Wallis test.

Results of Kruskal-Wallis tests adjusted for ties indicated that attitudes toward the hospice philosophy of care ( ${\chi }_2^2$  = 11.54, P = .003), religiosity ( ${\chi }_2^2$  = 96.51, P < .001), and years working in hospice ( ${\chi }_2^2$  = 23.88, P < .001) significantly differed by levels of MAID attitudes (see Table 3). Dunn’s post hoc tests showed that participants who opposed MAID reported significantly higher support for the hospice philosophy of care than those who neither supported nor opposed MAID (z = 3.1, P_BH = .003) and those who supported MAID (z = 2.6, P_BH = .008). Additionally, participants who opposed MAID were significantly more religious than those who neither supported nor opposed MAID (z = 4.2, P_BH < .001) and those who supported MAID (z = 9.8, P_BH < .001). Conversely, participants who neither supported nor opposed MAID were significantly more religious than those who were supportive (z = 2.7, P_BH = .003). Finally, participants who opposed MAID reported working in hospice significantly longer than those who neither supported nor opposed MAID (z = 3.0, P_BH = .002) and those who supported MAID (z = 4.8, P_BH < .001).

Multivariable

The partial proportional odds model was statistically significant (Wald ${\chi }_{18}^2$  = 265.65, P < .001, McFadden R² = .31, McFadden $R_{\mathrm{adj}}^2$  = .26; see Table 4). Results indicated that attitudes toward the hospice philosophy of care were not significantly associated with MAID attitudes at either threshold (above oppose: adjusted odds ratio [AOR] = 0.97, P = .391; above neither support nor oppose: AOR = 1.06, P = .095). However, compared to participants who agreed with the principle that hospice care should not hasten death, those who neither agreed nor disagreed (AOR = 5.92, P < .001) and those who disagreed altogether (AOR = 7.26, P < .001) showed roughly 6 and 7 times the cumulative odds of higher MAID support, respectively. Orientation toward patient-centeredness was not significantly associated with MAID attitudes (AOR = 1.32, P = .332). Relative to those with no professional exposure, participants exposed to working in states where MAID was legal had over twice the cumulative odds of higher MAID support (AOR = 2.36, P = .002).

Table 4.

Partial proportional odds model of attitudes toward medical aid in dying (N = 450).

Variable	Threshold 1: Above oppose (at or below oppose)				Threshold 2: Above neither support nor oppose (at or below neither support nor oppose)
	AOR (95% CI)	SE	z	P value	AOR (95% CI)	SE	z	P value
Constant	15.44 (0.39–608.24)	28.94	1.46	.144	0.22 (0.01–9.63)	0.43	−0.78	.435
Attitudes toward the hospice philosophy of care a	0.97 (0.91–1.04)	0.03	−0.86	.391	1.06 (0.99–1.14)	0.04	1.67	.095
Hospice care should not hasten death
Agree	1 [Reference]				1 [Reference]
Neither agree nor disagree	5.92 (3.40–10.30)	1.67	6.28	<.001	5.92 (3.40–10.30)	1.67	6.28	<.001
Disagree	7.26 (2.86–18.40)	3.45	4.18	<.001	7.26 (2.86–18.40)	3.45	4.18	<.001
Orientation toward patient-centeredness	1.32 (0.75–2.31)	0.38	0.97	.332	1.32 (0.75–2.31)	0.38	0.97	.332
Professional exposure to working in a state where medical aid in dying is legal
No	1 [Reference]				1 [Reference]
Yes	2.36 (1.36–4.07)	0.66	3.07	.002	2.36 (1.36–4.07)	0.66	3.07	.002
Age	1.00 (0.98–1.03)	0.01	0.31	.754	1.00 (0.98–1.03)	0.01	0.31	.754
Gender
Woman	1 [Reference]				1 [Reference]
Man	0.37 (0.22–0.61)	0.10	−3.83	<.001	0.37 (0.22–0.61)	0.10	−3.83	<.001
Religious identity
Protestant	1 [Reference]				1 [Reference]
Catholic	0.76 (0.37–1.58)	0.28	−0.73	.462	0.76 (0.37–1.58)	0.28	−0.73	.462
Other Christian	1.23 (0.65–2.32)	0.40	0.63	.530	1.23 (0.65–2.32)	0.40	0.63	.530
Unaffiliated	2.06 (0.83–5.12)	0.96	1.56	.118	2.06 (0.83–5.12)	0.96	1.56	.118
Other	1.19 (0.58–2.42)	0.43	0.47	.636	1.19 (0.58–2.42)	0.43	0.47	.636
Religiosity	0.81 (0.75–0.87)	0.03	−5.70	<.001	0.81 (0.75–0.87)	0.03	−5.70	<.001
Years working in hospice	0.97 (0.94–1.00)	0.02	−1.77	.077	0.97 (0.94–1.00)	0.02	−1.77	.077
Profession
Physician	1 [Reference]				1 [Reference]
Nurse a	1.90 (0.85–4.26)	0.78	1.56	.119	3.26 (1.49–7.13)	1.30	2.96	.003
Social worker	3.61 (1.71–7.61)	1.37	3.38	.001	3.61 (1.71–7.61)	1.37	3.38	.001
Chaplain	2.73 (1.40–5.33)	0.93	2.95	.003	2.73 (1.40–5.33)	0.93	2.95	.003

Wald ${\chi }_{18}^2$  = 265.65, P < .001, McFadden R² = .31, McFadden $R_{adj}^2$  = .26. Attitudes toward medical aid in dying were measured through a 3-point Likert-type item (0 = oppose, 1 = neither support nor oppose, 2 = support), resulting in the model’s imposition of two thresholds.

^aWe freed these variables from the proportional odds constraint^86,87 to model the variation in slopes across thresholds of attitudes toward medical aid in dying reflected in the results observed during assumption checks.

AOR, adjusted odds ratio.

Regarding covariates, men demonstrated 63% greater cumulative odds of MAID support than women (AOR = 0.37, P < .001). Although religious identity was not significantly associated (P > .05 for all), religiosity was. Specifically, each 1-point increase in religiosity was significantly associated with a decrease of 19% (AOR = 0.81, P < .001) in the cumulative odds of MAID support. Chaplains (AOR = 2.73, P = .003) and social workers (AOR = 3.61, P = .001) demonstrated around 3 and 4 times the cumulative odds of physicians in reporting higher MAID support, respectively. Although statistically nonsignificant when assessing the cumulative odds of being above oppose (AOR = 1.90, P = .119), nurses more than tripled physicians’ cumulative odds in being above neither support nor oppose (AOR = 3.26, P = .003). Neither age (AOR = 1.00, P = .754) nor years working in hospice (AOR = 0.97, P = .077) was significantly associated with MAID attitudes.

Discussion

Employing a national sample of interdisciplinary US hospice clinicians, this study appears to be the first to examine attitudes toward MAID vis-à-vis the hospice philosophy of care, patient-centered care, and professional exposure to working in a state where MAID is legal. Although results show that contextual factors, alongside individual characteristics, may shape MAID attitudes, mixed support across hypotheses highlights the complex associations between institutions and actors.

Professional exposure to working in a state where MAID is legal

In support of our hypothesis, we found that professional exposure to working in a state where MAID was legal was significantly associated with more supportive attitudes toward MAID. This finding extends previous research on a similar association considering the legal status of MAID where nurses lived ⁶² by showing that this result holds across samples of fully represented hospice interdisciplinary groups and by focusing, instead, on MAID legality where hospice patients receive care. Informed by institutional theory, ²³ more favorable MAID attitudes may reflect an actor’s increased socialization to institutional environments where MAID is clinically, politically, and socially legitimized. ⁶⁰ Indeed, declines in clinician apprehension following MAID implementation noted by members of California’s End of Life Option Act ⁸⁸ working group (C. L. Cain, oral communication, 24 February 2022) suggest that clinicians’ apprehension toward the prospect of MAID may exceed the negative impact of MAID’s implementation on their routine practice.^57,89 Consequently, personal opposition toward MAID may be predicated, in part and among other factors (eg, partisan sorting), ⁹⁰ on a lack of experience.

Attitudes toward the hospice philosophy of care

We found no multivariable support for our hypothesis that attitudes toward the hospice philosophy of care would be significantly associated with attitudes toward MAID. This finding is quizzical inasmuch as it contradicts our theoretically informed conceptualization of actors’ attitudes as expressions of values^26,27 advanced as institutional logics. ²⁴ It also contradicts previous research noting the hospice philosophy of care as driving institutional,^{34,36 –38} organizational,^{39 –41} and individual^42,43 MAID responses, both supportive and oppositional. Augmenting this perplexity is that our model accounted for the differing slopes in attitudes toward the hospice philosophy of care observed across thresholds of attitudes toward MAID. Despite aligning with previous research showing the hospice philosophy of care used to justify stances of both support and opposition,^{34,36,41,44 –47} modeling this difference through a partial proportional odds model precludes attributing this statistical nonsignificance to the linearity constraints imposed in traditional proportional odds models. Additional concerns related to the conflicting interpretations within^39,41,48 and among^{36,41,44 –46} principles included in the hospice philosophy of care are attenuated by recent psychometric support for the validity and reliability of the HPS-8 for use with hospice clinicians ⁷⁴ and the scale’s good ⁷⁶ internal consistency reliability demonstrated in the current study. These considerations in mind, reasons for this unanticipated result remain unclear and emphasize the need for further explication.

In lieu of any statistically significant association between attitudes toward the hospice philosophy of care and attitudes toward MAID, we found that MAID support was greater in participants who neither agreed nor disagreed or who disagreed with the principle that hospice care should not hasten death. In addition to aligning with previous research in terms of direction,^36,44,46 these cumulative odds ratios represented the strongest associations in the multivariable model. Although excluded from the National Hospice and Palliative Care Organization’s current iteration of the modern hospice philosophy of care ²² and the HPS-8 ⁷² used to measure hospice philosophy of care in the current study, this principle persists as a strong facet of hospice identity.^36,41,44,46 Against a backdrop of conflicting interpretations insofar as MAID,^{33,34,36,41,44 –46} the contrast between this statistically significant, inverse association and the statistically nonsignificant association with the hospice philosophy of care emphasizes the lack of any readily discernable relationship between the hospice institution and MAID.^41,91,92

Orientation toward patient-centeredness

Results also did not support the existence of a statistically significant association between orientation toward patient-centeredness and MAID attitudes, leading to the rejection of our final hypothesis. As with the hospice philosophy of care, this finding contests our theoretical conceptualization that patient-centered care, as a dominant institutional logic of hospice care, ²⁴ would underlie actors’ value-based appraisals of MAID attitudes.^26,27 The resulting lack of statistical significance contrasts with conceptual support,^54,55 as well as empirical support derived from qualitative studies referencing patient-centered reasons for clinicians’ support^{42,46,50,56,57} and previous quantitative studies finding positive correlations between key features of patient-centered care and MAID attitudes in clinicians.^14,58 Taken together, our results contribute to a research base reflecting an inconsistent understanding of how hospice clinicians perceive patient-centered care’s application in MAID. Given that patient-centered care is increasingly used as a quality metric in health care, ⁵² these results signal a need for further study.

Covariates

As potentially the most established correlate of MAID attitudes,^{8,10 –16} we found that increased religiosity was significantly associated with decreased MAID support. Although the cumulative odds ratio may appear modest, its value when distributed across the scale’s ⁸³ 15-point theoretical range represents a substantial cumulative effect. Other researchers have attributed similar results to religious values placed on the sanctity of life. ⁸ Thus, in light of religious identity’s statistical nonsignificance—itself potentially due to high representation in Christian religions—intensity of subscription to religious ideals may matter more for MAID attitudes than specific religious belief systems. Contradicting previous research,^{18 –20} physicians, not nurses, were the least supportive member of the hospice interdisciplinary group. This finding may be explained through physicians’ increased liability as MAID prescribers or differing interpretations of medical ethics. Anomalous results also extended to the finding that men in this sample were significantly less supportive of MAID than women^{8,10 –12} and that the associations with age^{8 –10} and years working in hospice ¹¹ and did not attain statistical significance. These equivocal results warrant additional investigation.

Limitations

Results should be considered in conjunction with study limitations. The nonprobability sampling approach through hospice and palliative care professional membership associations to which participants had to have limits on external validity. Therefore, results are generalizable neither to the membership of the professional membership associations sampled nor the broader population of hospice clinicians. Relatedly, professional membership associations’ inability to disaggregate nonhospice palliative care members from hospice care members precluded determining a precise response rate or differentiating responders from nonresponders to mitigate or otherwise characterize nonresponse bias.

One final limitation concerns potential measurement error in the PPOS ⁷⁷ and the HPS-8. ⁷² Despite the former’s recognition as the gold-standard measure of patient-centeredness, the current study coincides with previous research^75,77 finding suboptimal ⁷⁶ internal consistency reliability with health care professionals. The latter’s omission of the principle that hospice care does not hasten death highlights a dissensus regarding which specific principles underlie the hospice philosophy of care.^{36,41,44,46,72} Accordingly, although selected in the absence of alternatives, these associations should be interpreted only insofar as their measurement through these particular scales.

Implications

These results offer several implications for future research. Ostensibly most crucial is the need for additional study on the associations between the hospice philosophy of care, patient-centered care, and MAID. Qualitative research should elucidate the specific ways in which hospice clinicians do or do not perceive the hospice philosophy of care and patient-centered care as manifesting in MAID. Accounting for these developments, quantitative work should replicate and reappraise these associations following scale refinement within and beyond the United States. Additional future studies should expand the identification of contextual correlations of attitudes toward MAID to inform a more complete depiction of MAID attitude formation and change. Finally, future research should increase the explicit application and description of theory as a mechanism to guide the aforementioned efforts and streamline research around a common perspective.

Conclusion

As MAID legalization expands throughout the United States, elucidating correlates of hospice clinicians’ attitudes toward MAID will become an increasingly important component of advancing future practice and research. The current study tested three novel hypotheses. We found that professional exposure to working in a state where MAID is legal was significantly associated with more supportive attitudes toward MAID. Our orienting theory suggests that this finding may be attributable to different socializations to hospice care brought on by experience working in institutional environments that legitimize MAID. Conversely, we found no statistically significant associations between (1) attitudes toward the hospice philosophy of care and orientation toward patient-centered care—two oft-invoked foci of extensive and ongoing theoretical debate—and (2) attitudes toward MAID. These results provide an initial exploration of the role of contextual characteristics in hospice clinicians’ MAID attitudes.