Association of Race with End-of-Life Treatment Preferences in Older Adults with Cancer Receiving Outpatient Palliative Care

Abstract

Background: End-of-life discussions and documentation of preferences are especially important for older cancer patients who are at high risk of morbidity and mortality.

Objective: To evaluate influence of demographic factors such as religiosity, education, income, race, and ethnicity on treatment preferences for end-of-life care.

Methods: A retrospective observational study was performed on baseline data from a multisite randomized clinical trial of Dignity Therapy in 308 older cancer patients who were receiving outpatient palliative care (PC). Interviews addressed end-of-life treatment preferences, religion, religiosity and spirituality, and awareness of prognosis. End-of-life treatment preferences for care were examined, including preferences for general treatment, cardiopulmonary resuscitation (CPR), and mechanical ventilation (MV). Bivariate associations and multiple logistic regression analysis of treatment preferences with demographic and other baseline variables were conducted.

Results: Our regression models demonstrated that race was a significant predictor for CPR preference and preferences for MV, although not for general treatment goals. Minority patients were more likely to want CPR and MV than whites. Men were more likely to opt for MV, although not for CPR or overall aggressive treatment, than women. Higher level of education was a significant predictor for preferences for less aggressive care at the end-of-life but not for CPR or MV. Higher level of terminal illness awareness was also a significant predictor for preferences for CPR, but not MV or aggressive care at the end-of-life.

Discussion: Race was significantly associated with all three markers for aggressive care in bivariate analysis and with two out of three markers in multiple regression analysis, with minorities preferring aggressive care and whites preferring less aggressive care. Contrary to our hypothesis, income was not significantly associated with treatment preferences, whereas religion was significantly associated with all markers for aggressive care in bivariate models, but not in multiple regression models. Clinical Trial Registration Number NCT03209440.

Introduction

Patients with advanced cancers face considerable emotional, physical, and spiritual distress. The seminal SUPPORT study demonstrated that many patients have prolonged painful deaths often undergoing unwanted treatments and caregiver burden.¹ Many professional organizations, including the American Medical Association and the American Society of Clinical Oncology, have developed recommendations for improved care of patients with advanced cancer to help align treatments with patient preferences and values. The Institute of Medicine (IOM) has emphasized the importance of skill building for all health care providers in primary palliative care (PC).² In addition, the Commission on Cancer strongly recommends the availability of credentialed PC services for cancer patients with standards for the assessment and management of patient distress.^3–5

The Patient Self-Determination Act was passed by the US Congress in 1990. This legislation requires that federally funded health care institutions provide patients with information about their right to direct future medical care with advance directives (ADs) to be applied in the event of future decisional incapacity. ADs can promote early discussion about preferences surrounding death and dying, help lessen patients' anxiety about future decisions, guide decision making in the event that they lose decisional capacity and help reduce emotional burden on family caregivers.⁶ However, only 29% of United States adults had ADs according to a 2005 survey (Pew Research Center for the People and Press 2006) and even recent studies indicate that ADs may not be used,⁷ some call for rethinking ADs altogether.⁸ Others explored various barriers and solutions. Difficulties in applying previously stated wishes to actual subsequent circumstances have prompted suggestions to evaluate and codify personal dispositions that can guide families and practitioners,^9,10 as well as efforts to understand how decision making and adjustment to illness (by the patient and their family) interact.^11–13 Difficulties in communicating have led to studies of family communication^14–17 and online guides. Community norms have been addressed by local and state initiatives, conversation projects, including those that use lay navigators and address sociodemographic issues.^18–22 We think exploring the persistent gaps is important.

Certain cultural and socioeconomic factors are associated with a higher incidence of aggressive care at end of life and lower rates of advance care planning (ACP).¹¹ Factors including low education level, cultural factors in predominately black adults, and religiousness have been shown to be associated with preferences for more aggressive end-of-life care.²³ High religious coping in advanced cancer patients is associated with more intensive life-prolonging treatment near the end-of-life.²⁴ Black and Latino patients are less likely than white patients to have ADs, utilize hospice, and more likely to want aggressive care at the end-of-life.²⁵ Black patients were also more likely than white patients to have spiritual beliefs conflicting with PC goals, and express discomfort discussing death.²⁵

Several reasons are proposed to explain differences observed in end-of-life treatment preferences for care in minority (black, Hispanic) populations. Some factors may include disparities in access to health care, racism in health care (both explicit and implicit bias), institutional systemic racism, and patient mistrust of the health care system.²⁵ Both structural and institutional barriers to care exist for minority patients. These, coupled with systematic disadvantage due to United States history impact health information and ACP. However, the role of factors such as religiosity and socioeconomic factors such as education and income on treatment preferences at the end of life and attitudes toward ADs for ethnic and racial minorities are still poorly understood. The National Quality Forum and National Consensus Project for Quality Palliative Care identify provision of culturally sensitive care as one of eight core domains of high-quality PC and something all providers should strive to achieve.^26,27

ADs are especially important for older adults who are at high risk of morbidity and mortality, yet, many continue not to have ACP discussions.²⁸ ACP may give patients more control over the care that they receive.²⁹ By contrast, those who do not have ACP may receive unwanted higher intensity but poorer quality care, experience higher levels of pain and suffering, financial distress, and higher levels of family psychosocial distress.^30,31

Evaluating the role of sociodemographic factors affecting elder's attitudes toward ACP is important. If significant factors are identified, it could help guide efforts to develop additional culturally sensitive targeted initiatives to improve rates of ACP for ethnic and racial minorities and to lessen disparities in end-of-life care.

Objectives

In this multi-institutional observational study, we examine the treatment preferences and terminal illness awareness (TIA) of older cancer patients who were receiving outpatient PC services. Review of the literature demonstrates associations between demographic characteristics and treatment preferences. Our hypothesis predicts that our cancer patient population will be similar in comparison to the previously reported treatment preferences toward aggressive care in minority, low income, higher religiosity, and lower education level populations.

Methods

We utilized baseline data from 308 participants enrolled in a randomized clinical trial of Dignity Therapy.³² The study was approved by the IRBs of all participating sites.

Study participants were patients aged 55 years or older with a cancer diagnosis at six hospitals across the United States who were being seen in a PC ambulatory clinic. Participants were required to be English literate and physically able to complete the study (determined by a Palliative Performance Scale [PPS] score >50).^33–38 Exclusion criteria included being legally blind or cognitively impaired (determined by a Mini-Mental Status Examination [MMSE] score <24)³⁹ or history of psychosis determined by record review. Patients enrolled in another similar intervention study were excluded. In addition, efforts were made to recruit a sample where half of the participants were experiencing distress. Distress defined as a high score on either of two screening instruments, more specially, five or more problems rated ≥3 for Patient Dignity Inventory⁴⁰ and three or more problems rated ≥2 or at least one rated ≥3 for Brief Religious and Spiritual Struggles Scale (Brief RSS).⁴¹

Measures

Treatment preferences were measured with the three-item Hypothetical Advance Care Planning Scenario (H-CAP-S).⁴² For this standardized and well established^42,43 measure, the participant is presented with a scenario “If I have a terminal illness with weeks to live, and my mind is not working well enough to make decisions for myself … then my goals and specific wishes … would be”: The four response options for the general treatment preferences item were as follows (1) prolong life; treat everything, (2) attempt to cure, but reevaluate often, (3) limit to less invasive and less burdensome interventions, or (4) provide comfort care only. Using the same scenario, a second question identified preference for cardiopulmonary resuscitation (CPR) with three response options: (1) I want, (2) I am undecided, and (3) I do not want. A third question identified preference for mechanical ventilation (MV) with response options: (1) I want, (2) I want treatment tried. If no clear improvement stop, (3) I am undecided, (4) I do not want. The responses for all three questions were treated as ordinal with lower values indicating preferences for more aggressive care. Three broad groups of possible predictors of treatment preferences were examined: (1) demographic factors, (2) religious/spiritual factors, and (3) awareness of prognosis.

Demographic variables

From interviews or medical record review, information was obtained about age, sex, ethnicity, race, marital status, education level, and income.

Religious/spiritual variable

Religious and spiritual involvement were measured with three standard items.⁴⁴ The first item described self-reported religious affiliation coded as (1) Christian, (2) other, and (3) none, which included spiritual but not religious. Participants also described their level of religiousness and spirituality with similar response choices for both items: (1) not at all religious (or spiritual), (2) slightly religious, (3) moderately religious, or (4) very religious. Responses to these two items were treated as ordinal.

Awareness of prognosis

Awareness of prognosis was measured with self-reported question “How would you describe your current health status?” Response options included the following: (1) “Relatively healthy,” (2) “Seriously but not terminally ill,” and (3) “Seriously and terminally ill.” The item was taken from a measure of peaceful awareness that was used in a study of 280 patients with advanced cancer.⁴⁵ The response to this item is treated as ordinal with higher score indicating awareness of the terminal nature of their illness.

Statistical analysis

Data extracted from a REDCap database were imported to statistical software R⁴⁶ for analysis. Descriptive statistics, including mean, standard deviation, frequency, percentage, and the amount of missing data, were obtained. Multiple imputation was used to process missing data.⁴⁷ To examine the association between the ordinal outcome variables and various predictors, we utilized ordinal logistic regression. Benjamini-Hochberg procedure was used to adjust the observed significance levels of bivariate associations.⁴⁸ Statistically significance was set at p < 0.05.

Results

Demographic characteristics are presented in Table 1. Seventy-one percent were white, 22% were black, 7% were other minorities, and 6% were Latino. The majority (75%) had some college or more education. Most (70%) reported themselves to be moderately or very religious and 83% were moderately or very spiritual. Eighty-five percent minorities and 74% whites were Christian, although the difference was not statistically significant in this sample (p = 0.06). Minorities were significantly more religious (p < 0.001). By self-report, 36% of PC patients with a cancer diagnosis described themselves as relatively healthy, 34% described themselves as being seriously but not terminally ill, and 29% described themselves as being terminally ill. Treatment preferences are presented in Table 2. Approximately half (52%) reported a preference for less invasive treatments at the end-of-life or comfort care, 50% prefer to forgo CPR, and 51% prefer no MV.

Table 1.

Sample Characteristics (N = 308)

Variable	Missing	Category	Value
Age	0	Mean (SD)	66.4 (7.5)
Gender	1	Female (%)	171 (56)
		Male (%)	136 (44)
Race	1	White (%)	219 (71)
		Other or unknown (%)	88 (29)
Ethnicity	1	Hispanic or Latino (%)	17 (6)
		Not Hispanic or Latino (%)	290 (94)
Marital status	31	Married/partnered (%)	162 (58)
		Single (%)	115 (42)
Education	31	High school or less (%)	70 (25)
		Some college/vocational (%)	84 (30)
		Bachelor degree (%)	61 (22)
		Graduate degree (%)	62 (22)
Income	53	less than $10,000 (%)	27 (11)
		$10,000 to $19,999 (%)	36 (14)
		$20,000 to $29,999 (%)	29 (11)
		$30,000 to $39,999 (%)	19 (7)
		$40,000 to $49,999 (%)	16 (6)
		$50,000 or more (%)	128 (50)
Cancer stage	13	I (%)	29 (10)
		II (%)	19 (6)
		III (%)	57 (19)
		IV (%)	186 (63)
		Not staged (%)	4 (1)
Religious affiliation	26	Christian (%)	218 (77)
		Other religion (%)	33 (12)
		No religion (includes spiritual not religious) (%)	31 (11)
How religious are you	28	Not at all religious (%)	41 (15)
		Slightly religious (%)	43 (15)
		Moderately religious (%)	115 (41)
		Very religious (%)	81 (29)
How spiritual are you	26	Not spiritual at all (%)	10 (4)
		Slightly spiritual (%)	38 (13)
		Moderately spiritual (%)	100 (35)
		Very spiritual (%)	134 (48)
TIA	2	Relatively healthy (%)	111 (36)
		Seriously but not terminally ill (%)	105 (34)
		Seriously and terminally ill (%)	90 (29)

Other Race includes black or African American 69 (22%), Asian 4 (1%), American Indian or Alaska Native 1 (<1%), Native Hawaiian or other Pacific Islander 1 (<1%), Multiracial 3 (1%), and other or unknown 10 (3%).

Other religion includes Jewish 9 (3%), Buddhist 4 (1%), and other 20 (7%).

SD, standard deviation; TIA, terminal illness awareness.

Table 2.

Treatment Preferences

Variable	Missing	Category	No. (%)
Treatment preferences	6	Prolong life; treat everything	44 (15)
		Attempt to cure, but reevaluate often	101 (33)
		Limit to less invasive and less burdensome interventions	57 (19)
		Provide comfort care only	100 (33)
Cardiopulmonary resuscitation	2	I want	96 (31)
		I am undecided	58 (19)
		I do not want	152 (50)
Mechanical ventilation	1	I want	22 (7)
		I want treatment tried. If no clear improvement, stop	90 (29)
		I am undecided	39 (13)
		I do not want	156 (51)

Bivariate associations with treatment preferences are presented in Table 3. White patients were more likely to want less invasive or comfort care measures at the end of life (p = 0.01). Preference for more aggressive treatment was associated with education (p = 0.001), religion (p = 0.001), and religiosity (p = 0.004). Higher levels of TIA were associated with being more likely to choose less invasive interventions or comfort care (p = 0.03).

Table 3.

Bivariate Associations with Treatment Preferences

Variable	Category	Treat everything	Reevaluate often	Less invasive	Comfort only	p
Age	Mean (SD)	65.2 (7.3)	66.0 (7.5)	66.7 (6.5)	67.3 (8.1)	0.14
Gender	Female (%)	16	28	21	34	0.74
	Male (%)	13	40	15	32
Race	White (%)	9	35	23	34	0.01
	Other (%)	29	30	9	31
Ethnicity	Not Hispanic (%)	14	34	18	34	0.44
	Hispanic (%)	26	23	31	21
Marital	Single (%)	18	34	16	31	0.32
	Married or partnered (%)	12	33	21	34
Education	High school or less (%)	28	34	10	28	0.001
	Some college (%)	12	42	15	30
	Bachelor degree (%)	10	37	27	26
	Graduate degree (%)	7	19	26	49
Income	Less than $10,000 (%)	19	28	11	42	0.32
	$10,000–$19,999 (%)	22	32	20	26
	$20,000–$29,999 (%)	20	36	8	36
	$30,000–$39,999 (%)	20	37	24	20
	$40,000–$49,999 (%)	18	31	7	43
	$50,000 or more (%)	9	34	23	33
Religious Affiliation	Christian (%)	17	39	15	29	0.001
	Other (%)	12	10	32	46
	No religion (%)	3	21	30	46
How religious	Not at all (%)	4	22	25	48	0.004
	Slightly (%)	11	37	18	33
	Moderately (%)	15	38	16	30
	Very (%)	23	31	19	28
How spiritual	Not at all (%)	20	20	21	40	0.74
	Slightly (%)	4	49	17	30
	Moderately (%)	13	43	15	29
	Very (%)	18	23	22	36
TIA	Relatively healthy (%)	22	33	17	29	0.03
	Seriously but not terminally ill (%)	8	40	24	28
	Seriously and terminally ill (%)	13	27	16	44

Bivariate Associations with Treatment Preferences were statistically significant for variables with bolded p values.

Bivariate associations with CPR preferences are presented in Table 4. White patients were less likely to want CPR (p < 0.001) than minority patients. Religious affiliation was associated with preference for CPR (p = 0.03). Patients with higher levels of TIA were less likely to prefer CPR (p = 0.02).

Table 4.

Bivariate Association with Cardiopulmonary Resuscitation Preferences

Variable	Category	Yes	Undecided	No	p
Age	Mean (SD)	65.5 (7.0)	66.0 (7.3)	67.2 (7.8)	0.12
Gender	Female (%)	28	21	52	0.31
	Male (%)	36	17	47
Race	White (%)	24	19	56	<0.001
	Other (%)	49	18	33
Ethnicity	Not Hispanic or Latino (%)	31	20	50	0.85
	Hispanic or Latino (%)	41	6	53
Marital	Single (%)	31	20	48	0.85
	Married/partnered (%)	31	18	51
Education	High school or less (%)	34	21	45	0.08
	Some college/vocational (%)	35	20	44
	Bachelor degree (%)	32	20	48
	Graduate degree (%)	22	13	64
Income	less than $10,000 (%)	29	15	56	0.12
	$10,000 to $19,999 (%)	41	19	40
	$20,000 to $29,999 (%)	42	17	41
	$30,000 to $39,999 (%)	35	30	35
	$40,000 to $49,999 (%)	47	14	39
	$50,000 or more (%)	24	19	57
Religious Affiliation	Christian (%)	34	21	45	0.03
	Other religion (%)	27	13	61
	No religion (%)	18	12	70
How religious	Not at all religious (%)	21	17	62	0.06
	Slightly religious (%)	35	14	51
	Moderately religious (%)	28	22	50
	Very religious (%)	40	19	41
How spiritual	Not spiritual at all (%)	38	2	60	0.81
	Slightly spiritual (%)	26	21	54
	Moderately spiritual (%)	30	24	46
	Very spiritual (%)	33	16	51
TIA	Relatively healthy (%)	41	15	44	0.02
	Seriously but not terminally ill (%)	30	24	46
	Seriously and terminally (%)	21	18	61

Bivariate Associations with Cardiopulmonary Resuscitation Preferences were statistically significant for variables with bolded p values.

Bivariate associations with preferences for MV are presented in Table 5. White patients were less likely to want MV than minority patients (p < 0.001). Preference for MV or a trial of MV was also associated with religious association (p = 0.004) and religiosity (p = 0.004).

Table 5.

Bivariate Association with Preferences for Mechanical Ventilation

Variable	Category	Yes	Want tried	Undecided	No	p
Age	Mean (SD)	64.6 (5.5)	67.7 (7.8)	66.3 (8.7)	66.0 (7.1)	0.52
Gender	Female (%)	6	26	15	54	0.19
	Male (%)	9	34	10	47
Race	White (%)	4	27	12	57	<0.001
	Other (%)	16	36	14	34
Ethnicity	Not Hispanic (%)	8	29	12	51	0.54
	Hispanic (%)	0	29	18	53
Marital	Single (%)	10	28	16	46	0.30
	Married/partnered (%)	5	31	10	54
Education	High school or less (%)	14	22	14	51	0.07
	Some college/vocational (%)	8	39	13	40
	Bachelor degree (%)	3	29	11	56
	Graduate degree (%)	2	26	13	60
Income	Less than $10,000 (%)	17	14	10	59	0.68
	$10,000 to $19,999 (%)	9	21	20	51
	$20,000 to $29,999 (%)	10	27	20	43
	$30,000 to $39,999 (%)	14	40	11	35
	$40,000 to $49,999 (%)	17	20	7	55
	$50,000 or more (%)	2	35	11	52
Religion affiliation	Christian (%)	9	33	13	45	0.004
	Other religion (%)	3	16	17	64
	No religion (%)	1	23	5	72
How religious	Not at all religious (%)	0	25	8	66	0.004
	Slightly religious (%)	3	27	14	56
	Moderately religious (%)	10	28	11	51
	Very religious (%)	10	35	17	38
How spiritual	Not spiritual at all (%)	0	39	9	52	0.30
	Slightly spiritual (%)	7	17	12	64
	Moderately spiritual (%)	7	30	15	48
	Very spiritual (%)	8	31	12	49
TIA	Relatively healthy (%)	10	33	10	47	0.07
	Seriously but not terminally ill (%)	5	35	13	47
	Seriously and terminally (%)	7	18	15	60

Bivariate Associations with Preferences for Mechanical Ventilation were statistically significant for variables with bolded p values.

Multiple logistic regression models are presented in Table 6, where positive coefficients indicate preference for less aggressive treatments. Race was a significant predictor for CPR preference and preferences for MV, but not for the overall treatment goals. Minority patients were more likely to want CPR and MV treatment than white patients. Higher level of education was a significant predictor for preferences for less aggressive care at the end of life but not for preferences for CPR or MV. Higher level of TIA was also a significant predictor for preferences to forgo CPR but not for preferences for MV or for less aggressive care at the end of life. Interestingly, gender, while not significantly associated with any treatment preferences in bivariate analysis, was associated with preference for MV in the multiple regression model. Men were more likely to opt for MV than women. Age, ethnicity, religion, and degree of religion were not significantly predictive of preferences for end-of-life treatment, CPR preferences, or preference for MV in multiple regression models.

Table 6.

Multivariable Predictors of Treatment Preferences

Variable: (Reference group)	Treatment goals	CPR	MV
	Estimate (SE)	Estimate (SE)	Estimate (SE)
Age	0.022 (0.015)	0.024 (0.016)	−0.016 (0.015)
Gender: male (female)	−0.221 (0.233)	−0.363 (0.243)	−0.620 (0.245)*
Race: other (white)	−0.494 (0.270)	−1.004 (0.272)***	−1.003 (0.268)***
Ethnicity: Hispanic or Latino (non-Hispanic)	−0.434 (0.521)	0.159 (0.530)	0.209 (0.504)
Education	0.334 (0.121)**	0.052 (0.124)	0.119 (0.125)
Marital: married/partnered (single)	0.363 (0.282)	−0.138 (0.287)	0.423 (0.285)
Income	−0.090 (0.080)	0.038 (0.081)	−0.107 (0.082)
Religion: other religion (Christian)	0.530 (0.395)	0.423 (0.416)	0.732 (0.408)
Religion: no religion (Christian)	0.600 (0.441)	0.961 (0.500)	0.695 (0.501)
How religious	−0.167 (0.159)	0.040 (0.166)	−0.124 (0.167)
How spiritual	0.201 (0.158)	0.083 (0.172)	−0.022 (0.173)
TIA	0.251 (0.140)	0.355 (0.146)*	0.193 (0.146)

The coefficients are log odds ratios. Positive (respectively negative) coefficients indicate preference for less (respectively more) aggressive treatments.

^*p < 0.05, ^**p < 0.01, ^***p < 0.001.

CPR, cardiopulmonary resuscitation; SE, standard error.

Discussion

We examined six demographic factors considered important in end-of-life decision making, including (1) race, (2) religion, (3) gender, (4) income (5) education, and (6) TIA and their association with three markers for treatment preferences: (1) overall treatment preferences, (2) CPR, and (3) MV. The sample of older cancer patients receiving outpatient PC in six sites across the United States was evenly divided in their preference for aggressive care; about half want CPR and MV. Race was significantly associated with all three markers for aggressive care in bivariate analysis and with two out of three in the multiple regression models examined, with minorities preferring aggressive care and white patients preferring less aggressive care. Gender was not significantly associated with any marker in bivariate analysis, but men were significantly more likely than women to choose MV when adjusting for other predictors. Religion, religiosity, and education were significantly associated with some markers for aggressive care in some bivariate models, but in contrast to our original hypotheses, none was significantly associated with the outcomes when adjusting for other predictors. TIA was not significantly associated with preference for MV in any of the models but was significantly associated with treatment preferences in the bivariate analysis and with CPR in the bivariate and multiple regression analysis. These findings are similar to observations in clinical practice, where we see that on occasion patients may forgo aggressive treatments (i.e., chemotherapy near end of life) but still desire CPR and MV.

In multiple regression models, race was a significant predictor for two of three measures of treatment preferences, including CPR and MV and with a trend to significance for treatment goals (p = 0.07). This was in contrast to the literature and our own hypothesis. The unique aspect of this study's multivariate analysis of demographic factors is the identification of race itself as a factor in predicting treatment preferences, independent of income, religiosity, education, and so on. Minorities, who in our sample were predominately black, preferred aggressive care more than white patients. Prior research demonstrates that black patients prefer (1) aggressive care at end of life, (2) are more religious and use religion for coping at end of life, (3) are less likely to use hospice, and (4) more likely to prefer CPR and aggressive care.²⁵ Multiple factors could account for these findings. Racial disparities exist on the health care side due to disparities in care access, insurance, health provider explicit and implicit bias, and systemic institutional racism.^49,50 Distrust of the health care system for black patients also relates to the historical experience of being the subjects of medical experiments without having the opportunity to provide informed consent. Disparities exist from a patient perspective due to patient autonomy, lack of trust of the health care system, education as a confounding variable, and a cultural belief that choosing less aggressive care at end of life may be immoral.^25,51,52 Minorities might hold more fatalistic views and misconceptions regarding care, are more likely to present at advanced stages of disease, and less likely to receive cancer directed therapy citing concern that surgery may cause the cancer to spread.³⁵ Socially mediated expectations and personal dispositions certainly have a role^9–13 but are harder to study or predict with the research methods that are generally used.

In prior studies, religious coping in advanced cancer patients is associated with receipt of intensive life-prolonging medical care near death.¹³ Findings in prior studies are mixed and suggest that religious involvement is associated with preferences for more aggressive care, with a number of null findings especially if multivariate over bivariate analysis was applied that accounted for differences in ethnicity and education.^52,53 In a secondary analysis by Delgado-Guay et al.,³⁷ of a randomized controlled trial that evaluated the influence of physician communication styles on resuscitation preferences of advanced cancer patients, there was no significant association among the intensity of patients' religiosity and do-not-resuscitate preferences for either the patient in the video with simulated scenario in the study or the patients themselves. Patients with advanced cancer that report high use of positive religious coping strategies might have higher receipt of intensive life-prolonging medical care near death.²⁴ In another study, patients highly supported by religious communities who were also provided spiritual care from their medical team had greater odds of receiving hospice care and had lower odds of receiving aggressive end-of-life interventions or dying in an ICU.³⁸ These findings highlight the importance of the medical community providing spiritual support from Clinical Pastoral Educated (CPE) trained chaplains in hospital settings with high health literacy and good communication skills.

Differences in the findings of prior studies and ours may be related more to the influence of race over religion, in the association with the preference for aggressive care. In our sample, minority patients reported being significantly more religious than white patients. Second, our use of multivariate analysis instead of bivariate analysis may contribute to differences in study findings, especially given our large sample with generous representation of Minorities. More studies are needed to understand this complex relationship.

As highlighted above, despite our hypothesis, income level and education were not related to the consistent desire for more aggressive end-of-life treatment. Analysis demonstrated that patients with low education prefer aggressive care, but in contrast do not prefer CPR or MV, which is a form of aggressive care. Prior research demonstrates the more educated an individual, the less likely they are to want aggressive, often futile, care at end of life.⁵³

We defined TIA as an independent factor to predict the level of aggressiveness in care at the end-of-life. Our findings confirmed the hypothesis—the more aware patients are that their disease is terminal, the less likely they will desire aggressive end-of-life care—for only one of the three measures of aggressive end-of-life care. This finding was only the case for CPR.

Our study is limited to older cancer patients receiving outpatient PC and may not be generalizable to the population at large. Latino and Asian patients were underrepresented in our sample perhaps because of the study requirement that patients be English speaking. Furthermore, the vast majority of patients with a religious affiliation were Christian limiting our ability to test our hypothesis for other religions. More research is needed to evaluate different cultures in different settings of care. Medical teams should consider providing spiritual care to patients and families, even those receiving spiritual care from the community.

Our study findings emphasize the importance of racial differences and disparities in clinical practice. Shared and informed decision making, good communication, and culturally sensitive care increase the likelihood of goal concordant care. One study demonstrated that 75% of the patients believed they understood the meaning of “DNR,” but only 32% were able to provide an accurate definition.⁵² Those who lacked an accurate understanding of DNR status were significantly more likely to perceive DNR as morally wrong.⁵²

Going forward, it is critically important for health care providers and institutions to recognize, own, and aim to address systemic racism and health care disparities that exist to improve end-of-life care. Standardized medical education that incorporates implicit bias education and competency, increasing awareness of racism, and providing evidence-based strategies can help break the bias habit.⁵⁰ Being sensitive to the types of bias that can occur is a key component to recognizing and overcoming systemic racism (recognize, name, challenge stereotypes, aim to improve the status quo through identification of variability among minority groups and trying to gain perspective of patients beliefs, and values).⁵⁰ PC specialty education should incorporate cultural/religious humility training, include bias reduction training, and strive for work force diversity that represents patient diversity. Such training should emphasize the importance of involvement of family and faith community and the development of curiosity about the lived experience differences and respect for minority patients.⁵⁴ Future research goals should define race, name racism, avoid blaming the patient or attributing differences to genetics variables, and seek patient input to help close this gap.⁴⁹

Conclusion

Our findings reflect the growing national focus on delivering goal-concordant cancer care. In a sample of cancer patients receiving outpatient concurrent PC, one-third perceived themselves as healthy and 15% preferred aggressive care. From six factors—sex, race, religion, income, education, and TIA, we found that race was significantly and independently associated with two of three markers for aggressive care—CPR and MV. Our study findings highlight the importance of an integrated culturally competent environment to provide the best quality of care to patients with advanced and terminal illness

Disclaimer

The information in this article is solely the responsibility of the authors and does not necessarily represent the views of the NIH, NCI, or NINR. The final peer-reviewed article is subject to the National Institutes of Health Public Access Policy.

Funding Information

This research was made possible by Grant Number 1R01CA200867 from the National Institutes of Health (NIH), National Cancer Institute (NCI), and National Institute of Nursing Research (NINR).

Author Disclosure Statement

No competing financial interests exist.