Characterizing Interdisciplinary U.S. Hospice Clinician Presence During Patient Self-Administration of Medical Aid in Dying Medication

Abstract

Background:

Despite emerging as a clinical and scholarly focus in medical aid in dying (MAID)-related care, limited data exist to characterize hospice clinician presence while patients self-administer the medication to hasten their death.

Objectives:

To explore (1) the proportion and (2) correlates of interdisciplinary hospice clinician presence during patient self-administration of MAID medication.

Design:

Exploratory secondary analysis of cross-sectional survey data.

Setting/Subjects:

Convenience sample of interdisciplinary U.S. hospice clinicians reporting permissive state and organizational MAID policy.

Measurements:

We assessed the proportion of the sample ever having been present during patient self-administration of MAID medication via frequency and percentage. We examined personal, professional, organizational, and MAID-specific characteristics as correlates via multiple logistic regression analysis adjusted for small sample bias.

Results:

Our sample included 100 hospice physicians, nurses, social workers, and chaplains. Descriptive results revealed that just over one-third of the sample had ever been present during patient self-administration of MAID medication. Regression results indicated that being a chaplain and working for a hospice with a policy permitting full MAID participation were each significantly associated with greater odds of ever having been present. Conversely, never having provided end-of-life care beyond information provision related to a hospice patient’s use of MAID was significantly associated with lower odds of ever having been present.

Conclusions:

Hospice clinician presence during patient self-administration of MAID medication appears relatively common and related to select professional, organizational, and MAID-specific characteristics. Improved annual state reporting practices and expanded replication efforts are warranted.

Introduction

Medical aid in dying (MAID) describes the process whereby a qualified individual self-administers medications prescribed by an attending provider that will result in hastened death.¹ As of this writing, MAID is legal in 10 U.S. states and the District of Columbia (hereinafter, “states”), accounting for ≈75 million adults, or one-fifth of the U.S. population.² Despite >100 years of combined implementation experience across these states, there remains a lack of knowledge about interdisciplinary hospice clinicians’ participation in MAID,^3–11 especially beyond minimum legal requirements (viz., patient education, eligibility assessment, prescription writing).¹ This gap is concerning because statutory protections for conscientious objection¹ mean that clinician participation may facilitate or obstruct patient receipt of lawful, goal-concordant care at the end of life.^12–14 Informed by sustained increases in annual utilization¹⁵ and presence on state ballot measures,^1,16 MAID’s growing presence in clinical care underscores the need to elucidate hospice clinician participation.^{4,6,7,17–24}

One area gaining increasing clinical and scholarly attention is clinician presence while patients self-administer the MAID medication,^4,25–28 which may occur via enteral, rectal, or, most commonly, oral routes.¹ Available data have stemmed almost exclusively from studies about broader clinician MAID participation.^{11,14,17,22,24,29–32} These studies have shown that clinician presence during this “critical”²⁸ and defining juncture of MAID delivery may reflect patient care preferences.^22,24,33 Primary reasons for these requests have included the desire for clinical support should postingestion complications arise^24,33 and for general psychosocial support.²² Additional studies have indicated that some hospice clinicians share the desire to be present.^4,31,32 Indeed, the sole study focused exclusively on hospice clinician presence found that just under three-quarters of an interdisciplinary national convenience sample reported hypothetical willingness to be present, with qualitative rationales premised on personal, clinical, and professional values.⁴ In practice, clinicians who have been present have reported providing key medical (e.g., procedural education, medication and supply arrangement, symptom monitoring) and psychosocial (e.g., bereavement services coordination) support.^{4,14,21,22,33} These findings suggest that both patients and hospice clinicians perceive clinical benefits conferred through clinician presence.

Despite these stakeholder-perceived benefits, empirical data on hospice clinician presence remain scant. This scarcity may stem, in part, from hospice policies that discourage clinician presence, even in legalizing states.^{18–20,29,34,35} All MAID statutes specify that clinician participation, as outlined therein, does not violate the law. Nevertheless, concerns about clinician participation—or speculated participation—in activities not permissible under MAID (e.g., clinician administration⁴) may compel hospices to implement policies that prohibit or restrict clinician presence.^{25,27,28,35,36} These restrictions often condition where clinicians may be physically located throughout the procedure.^{25,27,28,35,36} Furthermore, even when hospice clinicians are present, data are limited by current state reporting practices. The noted lack of minimum reporting standards across annual utilization reports^1,37 is exemplified in only California³⁸ and Oregon³⁹ reporting on clinician presence. Moreover, although both states identify clinician attendees’ roles in the context of MAID (e.g., attending provider, consulting provider, other [e.g., volunteer]), these reports do not indicate whether or not those in attendance are affiliated with hospice care. Thus, restrictive hospice policies may contribute to an underrepresentation of hospice clinicians during patient self-administration of MAID medication, while downstream, epidemiological data on hospice clinician presence is obscured by limitations in state reporting.

Stakeholder support, alongside the lack of empirical data, emphasizes emerging calls to characterize hospice clinician presence, while patients self-administer MAID medication to hasten death.^4,25–28 In response, the objectives of the current study were to explore (1) the proportion and (2) correlates of interdisciplinary hospice clinician presence during patient self-administration of MAID medication.

Methods

Study design

This exploratory study used secondary data from our broader, cross-sectional study on hospice clinicians’ attitudes toward MAID.^3,4,40 For the broader study, we recruited a convenience sample of interdisciplinary hospice clinicians across the United States. Eligible participants had to be ≥18 years old, work as a paid hospice employee, and provide direct patient care. Participants were recruited through the membership lists of prominent hospice and palliative care professional membership associations to complete a self-administered, one-time Qualtrics survey. Each association (unnamed per research agreements) represented one of the four disciplines constituting the Medicare hospice benefit defined “hospice interdisciplinary group” (medicine, nursing, social work, and spiritual care).⁴¹ Maximizing the terms in each research agreement prompted variation in the number of survey contacts (2 vs. 5), contact source (association vs. research team), and contact mode (e-mail vs. newsletter). We defined survey completion as responses to at least half of all survey items.⁴² Surveys remained open for 30 days, spanning November 2022 through January 2023.

Despite implementing expert-recommended⁴³ survey security features in Qualtrics,⁴⁴ we suspected bot infiltration of our second nursing survey, after receiving an unusually high number of responses within 2 days (n = 2392). Thus, we list-wise deleted each nursing case recorded following this dissemination. Our compensatory wave of data collection produced no concerns of fraudulent response. We compensated 200 randomly selected participants with $20 egift cards. The Washington University Institutional Review Board approved this study as nonhuman subjects’ research.

Measures

Data were collected through a 68-item self-report survey about hospice clinicians’ attitudes and experiences with MAID.⁴⁰ Following review of the prospective survey by an expert panel (N = 5) of aging and end-of-life care researchers with clinical practice experience for content validity and face validity,⁴⁵ the first author (T.D.B.) conducted cognitive interviewing (N = 8) and pilot testing (N = 11) of the refined survey with typical-case purposive samples of hospice clinicians to assess item construction and technological functionality, respectively.⁴⁵ These procedures informed several minor revisions to survey items for improved clarity.

Hospice clinician presence during patient self-administration of MAID medication.

Hospice clinician presence during patient self-administration of MAID medication was assessed through a single item asking if participants had ever been present while a hospice patient self-administered MAID medication (yes, no).

Personal characteristics.

Personal characteristics included age in years, gender (man, woman), race (White, person of color), religious identity (Christian, agnostic/atheist, other), and religiosity. Religiosity was assessed through the five-item version of the Santa Clara Strength of Religious Faith Questionnaire.⁴⁶ Items are scored from 1 (strongly disagree) to 4 (strongly agree). Higher sum scores indicate higher religiosity (theoretical range = 5–20). Previous research has supported internal consistency reliability,^3,47,48 test-retest reliability,⁴⁷ and convergent validity.^47,48 This measure demonstrated excellent⁴⁹ internal consistency reliability in the current study (Cronbach α = .95).

Professional characteristics.

Professional characteristics included professional discipline (physician, nurse [registered nurse/advanced practice registered nurse], social worker, chaplain) and duration of time working in hospice in years.

Organizational characteristics.

Organizational characteristics included approximate average daily patient census, tax status (for profit, not for profit), religious affiliation (yes, no), and policy on permitted MAID participation (full, partial).

MAID-specific characteristics.

MAID-specific characteristics included agreement with the principle that hospice care should not hasten death (agree, neither agree nor disagree, disagree); general attitude toward MAID (support, neither support nor oppose, oppose); approximate number of hospice patients who had ever inquired about MAID; whether or not participants had ever provided information about MAID to a hospice patient (yes, no); whether or not patients had ever provided end-of-life care beyond information provision related to a hospice patient’s use of MAID (yes, no); and willingness to be present in the room with a hospice patient using MAID from self-administration until after death, provided hypothetical state legality, organizational permission, and voluntary/explicit patient request (yes, no/unsure).⁴

Statistical analysis

We characterized the sample via descriptive statistics. We assessed the proportion of the sample present during patient self-administration of MAID medication by way of frequency and percentage. We examined correlates first through chi-square tests, Fisher’s exact tests, and Wilcoxon rank-sum tests. Independent variables yielding statistically significant bivariable results were retained for multivariable analysis via multiple logistic regression. Due to concerns over small sample bias, we applied Firth’s⁵⁰ penalized maximum likelihood estimator,⁵¹ using the user-written firthlogit⁵² command. We evaluated model fit, using the user-written firthfit⁵³ command. We conducted all statistical analyses in Stata (Version 19.5) and defined statistical significance (p < 0.05) in line with conventional thresholds (α = 0.05 [two-sided]).

Results

Analytic sample specification

We specified the analytic sample for this secondary analysis via a four-step process of exclusions from the 1346 total responses to the broader survey. First, we removed cases based on predetermined study design criteria (850 of 1346 [63.2%]: lack of informed consent, 130; study ineligibility, 683; survey break-off,⁴² 26; Qualtrics security metrics,⁴⁴ 11). Next, we deleted cases for whom we determined MAID presence would be implausible due to policy (388 of 496 [78.2%]: contemporaneous state MAID illegality within hospice service area, 362; hospice MAID policy not permitting at least partial employee participation, 26). Then, we excluded cases based on viability for analysis (2 of 108 [1.9%]: sparseness in variable categories, 2). Finally, we dropped cases containing missing data on variables indicated for analysis (6 of 106 [5.7%]). Missingness ranged from 1 (0.9%) on race, a religiosity item, and hospice religious affiliation to 3 (2.8%) on participant religious identity. We proceeded with complete case analysis after results of Little’s test⁵⁴ indicated that data were missing completely at random (χ²₁₂ = 13.2, p = 0.36).⁵⁵

Sample characteristics

The sample (N = 100) included 44 (44.0%) physicians, 23 (23.0%) chaplains, 21 (21.0%) social workers, and 12 (12.0%) nurses employed in hospices with service areas encompassing each state where MAID was legally available at the time of data collection (see Table 1). Participants were between 26 and 74 years of age (mean [standard deviation or SD] = 53.5 [12.2] years) and were mostly women (68 [68.0%]), White (92 [92.0%]), and Christian (46 [46.0%]). Duration of time working in hospice ranged from 1 through 40 years (mean [SD] = 11.3 [8.3] years). The hospices for which participants worked were mostly not for profit (75 [75.0%]) and religiously unaffiliated (79 [79.0%]).

Table 1.

Sample Characteristics (N = 100)

Characteristic	No. (%)
Participant
Age, mean (SD) [range], years	53.5 (12.2) [26.0–74.0]
Gender
Man	32 (32.0)
Woman	68 (68.0)
Race
White	92 (92.0)
Person of color	8 (8.0)
Religious identity
Agnostic/atheist	22 (22.0)
Christian	46 (46.0)
Other	32 (32.0)
Professional discipline
Chaplain	23 (23.0)
Nurse (RN/APRN)	12 (12.0)
Physician	44 (44.0)
Social worker	21 (21.0)
Duration of time working in hospice, mean (SD) [range], years	11.3 (8.3) [1.0–40.0]
Employing hospice
Tax status
For profit	25 (25.0)
Not for profit	75 (75.0)
Religious affiliation
Yes	21 (21.0)
No	79 (79.0)

APRN, advanced practice registered nurse; RN, registered nurse; SD, standard deviation.

Proportion of hospice clinician presence during patient self-administration of MAID medication

Descriptive results revealed that just over one-third of the sample had ever been present during patient self-administration of MAID medication (34 [34.0%]).

Correlates of hospice clinician presence during patient self-administration of MAID medication

Bivariable.

Chi-square test results indicated significantly higher frequencies of ever having been present during patient self-administration of MAID medication in participants working for hospices with policies permitting full MAID participation (full, 22 [62.9%] vs. partial, 12 [18.5%]; χ²₁ = 20.0, p < 0.001), participants who had ever provided end-of-life care beyond information provision related to a hospice patient’s use of MAID (yes, 30 [45.5%] vs. no, 4 [11.8%]; χ²₁ = 11.4, p = 0.001), and participants indicating hypothetical willingness to be present in the room with a hospice patient using MAID from self-administration until after death (yes, 32 [40.5%] vs. no/unsure, 2 [9.5%]; χ²₁ = 7.1, p = 0.01; see Table 2). Further results indicated the highest frequencies of presence during patient self-administration of MAID medication in chaplains (physician, 9 [20.5%]; nurse, 3 [25.0%]; social worker, 10 [47.6%]; chaplain, 12 [52.2%]; χ²₃ = 9.2, p = 0.03), participants who disagreed with the principle that hospice care should not hasten death (agree, 11 [22.0%]; neither agree nor disagree, 17 [44.7%]; disagree, 6 [50.0%]; χ²₂ = 6.5, p = 0.04), and participants whose general attitude toward MAID was supportive (support, 31 [40.8%]; neither support nor oppose, 1 [10.0%]; oppose, 2 [14.3%]; χ²₂ = 6.6, p = 0.04).

Table 2.

Bivariable Tests of Clinician Presence During Patient Self-Administration of Medical Aid in Dying Medication (N = 100)

	No. (%)
Variable	No	Yes	χ2 (df)	p Value
Personal characteristics
Age, yearsa	66 (3252.5)	34 (1797.5)	−0.6	0.56
Genderb			0.3 (1)	0.61
Man	20 (62.5)	12 (37.5)
Woman	46 (67.7)	22 (32.4)
Racec			NA	>0.99
White	61 (66.3)	31 (33.7)
Person of color	5 (62.5)	3 (37.5)
Religion			1.3 (2)	0.52
Agnostic/atheist	13 (59.1)	9 (40.9)
Christian	33 (71.7)	13 (28.3)
Other	20 (62.5)	12 (37.5)
Religiositya	66 (3490.5)	34 (1559.5)	1.2	0.25
Professional characteristics
Professional disciplineb			9.2 (3)	0.03
Physician	35 (79.6)	9 (20.5)
Nurse (RN/APRN)	9 (75.0)	3 (25.0)
Social worker	11 (52.4)	10 (47.6)
Chaplain	11 (47.8)	12 (52.2)
Duration of time working in hospice, yearsa	66 (3301.0)	34 (1749.0)	−0.2	0.82
Organizational characteristics
Approximate average daily patient censusa	66 (3218.5)	34 (1831.5)	−0.8	0.41
Tax status			0.6 (1)	0.81
For profit	17 (68.0)	8 (32.0)
Not for profit	49 (65.3)	26 (34.7)
Religious affiliationb			0.9 (1)	0.34
Yes	12 (57.1)	9 (42.9)
No	54 (68.4)	25 (31.7)
Policy on permitted MAID participation			20.0 (1)	<0.001
Full	13 (37.1)	22 (62.9)
Partial	53 (81.5)	12 (18.5)
MAID-specific characteristics
Agreement with the principle that hospice care should not hasten death			6.5 (2)	0.04
Agree	39 (78.0)	11 (22.0)
Neither agree nor disagree	21 (55.3)	17 (44.7)
Disagree	6 (50.0)	6 (50.0)
General attitude toward MAID			6.6 (2)	0.04
Support	45 (59.2)	31 (40.8)
Neither support nor oppose	9 (90.0)	1 (10.0)
Oppose	12 (85.7)	2 (14.3)
Approximate number of hospice patients who had ever inquired about MAID	66 (3156.5)	34 (1893.5)	−1.3	0.20
Ever provided information about MAIDc			NA	0.16
Yes	61 (64.2)	34 (35.8)
No	5 (100.0)	0 (0.0)
Ever provided end-of-life care related to hospice patient’s use of MAIDb			11.4 (1)	0.001
Yes	36 (54.6)	30 (45.5)
No	30 (88.2)	4 (11.8)
Hypothetical willingness to be present			7.1 (1)	0.01
Yes	47 (59.5)	32 (40.5)
No/unsure	19 (90.5)	2 (9.5)

^aStatistics reflect the Wilcoxon rank-sum tests and are presented as No. (rank-sum), No. (rank-sum), U, and p value, respectively.

^bBecause of rounding, percentages may not total 100.

^cStatistics reflect Fisher’s exact tests, which do not produce point estimates or corresponding degrees of freedom.

APRN, advanced practice registered nurse; MAID, medical aid in dying; NA, not applicable; RN, registered nurse.

Multivariable.

The resulting multiple logistic regression model provided a significantly better fit to the data than a null model (Wald χ²₁₂ = 22.3, p = 0.01, Tjur R² = 0.34; see Table 3). Compared with physicians, chaplains had >4 times greater odds of ever having been present during patient self-administration of MAID medication (adjusted odds ratio [AOR] = 4.32; 95% confidence interval [CI]: 1.22–15.38). Participants working for hospices with policies permitting full MAID participation demonstrated 4.5 times greater odds of ever having been present during patient self-administration of MAID medication, relative to those permitting only partial MAID participation (AOR = 4.50; 95% CI: 1.58–12.86). Participants who had never provided end-of-life care beyond information provision related to a hospice patient’s use of MAID demonstrated 74% lower odds of ever having been present during patient self-administration of MAID medication than those who had (AOR = 0.26; 95% CI: 0.67–0.98).

Table 3.

Multiple Logistic Regression Model of Clinician Presence During Patient Self-Administration of Medical Aid in Dying Medication (N = 100)^a

Variable	AOR (95% CI)	SE	z	p Value
Intercept	0.26 (0.08–0.86)	0.16	−2.21	0.03
Professional characteristics
Professional discipline
Physician	1 [Reference]	[Reference]	[Reference]	[Reference]
Nurse (RN/APRN)	1.63 (0.28–9.46)	1.46	0.55	0.58
Social worker	1.68 (0.46–6.20)	1.12	0.78	0.44
Chaplain	4.32 (1.22–15.38)	2.80	2.26	0.02
Organizational characteristics
Policy on permitted MAID participation
Partial	1 [Reference]	[Reference]	[Reference]	[Reference]
Full	4.50 (1.58–12.86)	2.41	2.81	0.01
MAID-specific characteristics
Agreement with the principle that hospice care should not hasten death
Agree	1 [Reference]	[Reference]	[Reference]	[Reference]
Neither agree nor disagree	1.14 (0.36–3.56)	0.66	0.23	0.82
Disagree	2.44 (0.54–11.11)	1.89	1.15	0.25
General attitude toward MAID
Support	1 [Reference]	[Reference]	[Reference]	[Reference]
Neither support nor oppose	0.33 (0.05–2.23)	0.32	−1.14	0.26
Oppose	1.07 (0.19–6.12)	0.95	0.08	0.94
Ever provided end-of-life care related to hospice patient’s use of MAID
Yes	1 [Reference]	[Reference]	[Reference]	[Reference]
No	0.26 (0.07–0.98)	0.18	−1.99	0.05
Hypothetical willingness to be present
Yes	1 [Reference]	[Reference]	[Reference]	[Reference]
No/unsure	0.48 (0.11–2.12)	0.36	−0.98	0.33

^aWald χ²₁₂ = 22.3, p = 0.01, Tjur R² = 0.34. Due to concerns over small sample bias, we applied Firth’s⁵⁰ penalized maximum likelihood estimator.

AOR, adjusted odds ratio; CI, confidence interval; SE, standard error.

Discussion

To our knowledge, this study is the first to explore the proportion and correlates of interdisciplinary hospice clinician presence during patient self-administration of MAID medication. Bolstered by data from participants working for hospices servicing each state where MAID is legal, results show that, when permitted by state and organizational policy, hospice clinicians participate in MAID-related care beyond minimum legal requirements.

Results from our first objective revealed that just over one-third of our sample had ever been present while a patient self-administered MAID medication. In light of restrictive organizational polices toward MAID,^{25,27,29,35,36} this result was higher than anticipated. It may be that, for various reasons, individual hospice clinicians are more amenable to MAID participation than the institutions and organizations for which they work.^4,32 Although our non-probability sampling approach did not equip us to estimate population prevalence, it bears mentioning that our sample proportion may not align with population prevalence. Nevertheless, considering upstream challenges in identifying MAID clinicians for research participation,⁵⁶ this result offers an initial step in said direction.

Results from our second objective indicated that select professional, organizational, and MAID-specific characteristics were significantly associated with hospice clinician presence during patient self-administration of MAID medication. We did not detect any statistically significant associations pertaining to personal characteristics. This finding may be due to a genuine lack of statistically significant association or insufficient power. Regardless, our results suggest that hospice clinicians’ personal characteristics are not primary drivers of their presence during patient self-administration of MAID medication.

In contrast, we found professional characteristics, namely professional discipline, to be significantly and strongly associated. Results showed that chaplains had considerably greater odds of ever having been present during patient self-administration of MAID medication than physicians. Alongside the general reality that many hospice physicians make few—if any—direct patient visits,⁵⁷ this finding may highlight variation in primary clinical MAID responsibilities by professional role.^6,9,10 Despite noted overlap,⁶ studies have suggested that physicians may primarily participate in the core steps marking the clinical pathway to patient prescription receipt, as defined in MAID statutes (viz, patient education, eligibility assessment, prescription writing).^6,8,24 Conversely, chaplains provide support predominantly tied to spiritual aspects of MAID,^11,58,59 which may be especially desirable around the time of self-administration. Therefore, provision of these distinct domains of support may culminate in time points of involvement that vary by professional discipline.^10,59 Notably, despite their provision of key clinical, psychosocial, and spiritual support resulting in consistent reports of presence,^9,10,21,22 nursing profession was not found to be significantly associated with presence. This unanticipated finding warrants further investigation, especially noting the sparseness of nurses in our sample.

Regarding organizational characteristics, we found that participants working for hospices with policies permitting full MAID participation demonstrated greater odds of presence during patient self-administration of MAID medication than those working for hospices permitting only partial participation. This finding is consistent with extant research, spotlighting organizational participation as a matter of potential liability.^18–20,34 In this sense, clinician-reported concerns about the optics of coercion or requests to administer MAID medication if present⁴ may be most expediently addressed through organizational policies that keep hospices and their staff “at arm’s length”²⁸ from the procedure. A recent national analysis of hospice organizational MAID participation policies found great heterogeneity not only across the specific domains of participation reported but also in the extent of participation permitted.²⁹ Our survey did not include measures clarifying which specific domains of participation were permitted or not at each participant’s hospice. However, it is likely that hospices would more readily prohibit domains of participation widely regarded as posing greater liability, such as prescription writing. Consequently, hospices permitting full participation may perceive limited additional risk in permitting lawful participation in other domains of patient preferences for MAID.

Considering MAID-specific characteristics, we found that having provided end-of-life care beyond information provision related to a hospice patient’s use of MAID (e.g., prescription writing) was significantly associated with greater odds of presence during patient self-administration of MAID medication than never having done so. As above, clinicians who participate in some domains of MAID may be more apt to participate in additional domains. It remains unclear, however, why other variables presumably reflecting this same rationale (e.g., provision of information about MAID) did not reach statistical significance.

Limitations

Results should be interpreted with limitations in mind. First, we recruited a convenience sample predicated on membership to select professional membership associations. Due to this sampling bias, these results do not generalize beyond our sample. Next, some partnering associations were unable to disaggregate nonhospice palliative care-based members from hospice-based members, precluding our ability to assess for nonresponse bias. Nevertheless, mode effects introduced through maximizing each research agreement’s terms to facilitate recruitment may have led to differences in response by association. Additionally, this secondary analysis may be limited by data availability. Given that these data were collected for a study with a broader MAID focus, the lack of measures clarifying the specific activities permitted across hospices with policies permitting only partial MAID participation may threaten internal validity. Nevertheless, this threat may be attenuated by our finding that nearly one-fifth of participants working for hospices with policies permitting only partial participation had ever been present. Last, although statistically mitigated to the extent possible in our multivariable model, our modest sample size may have contributed to type II error.

Implications

This study offers several implications for practice and research. Regarding the former, improved state reporting of MAID utilization is needed. The lack of standardization is exacerbated by our observation that only California³⁸ and Oregon³⁹ currently report on clinician presence during patient self-administration of MAID medication. Wider documentation of clinician presence and health care specialty would inform the development of tailored training opportunities and support services for the health care teams and personnel formatively involved in MAID.

Regarding the latter, future quantitative research with larger samples should replicate our results. Such efforts should build on our approach by accounting for the specific permissibility of clinician presence in hospice MAID participation policies. They should also include factors absent in our data set (e.g., types of MAID-related care previously provided). Additional investigations should expand our focus by considering clinicians’ physical locations at other time points throughout the MAID procedure (e.g., between self-administration and death, after death) to provide a more holistic understanding of hospice clinician participation.

Key Message.

We found that select professional, organizational, and medical aid in dying (MAID)-specific characteristics were related to hospice clinician presence during patient self-administration of MAID medication. Our results show that, when permitted, hospice clinicians participate in domains of MAID-related care beyond minimum legal requirements. Findings suggest need for broader annual state reporting and empirical replication.