The Impact of Race on Advanced Care Planning: Results of a Quality Improvement Project

Introduction

The goal of advanced care planning (ACP) is to ensure that a patient’s end-of-life wishes are discussed, documented, and upheld. End-of-life decisions are complex. In addition to the patient, decisions involve family members/surrogates and the healthcare team.¹ Benefits of ACP include improved end-of-life communication, increased satisfaction with end-of-life care, a perception of more control over care at the end of life for patients and families/surrogate decision-makers, care that is concordant with patient goals, and potentially decreased healthcare costs.^2,3

Medicare has created international classification of disease 10 (ICD-10) codes for ACP discussion with patients⁴ and set an expectation for ACP discussion as part of Medicare wellness examinations.⁵ Published research indicates a racial disparity in completion of ACP, with non-White patients less likely to engage in ACP.^2,6,7 The reasons for this disparity are not entirely clear. While it has been suggested and explored by researchers, the evidence is not conclusive as to whether a lack of trust in the medical system among minority patients plays a role in the disparity of ACP.^8,9 There is evidence to suggest that Black patients prefer a close relationship with caregivers in comparison to White patients at the end of life,¹⁰ suggesting that trust may play a role. Some authors have suggested that ACP does not change the likelihood of goal-concordant care for patients regardless of race.^1,11 Others have demonstrated that, at the end of life, Black patients appear to have more emergency department visits and hospitalizations and use hospice care less often.^12,13

As part of a larger project focused on equity within our institution,¹⁴ we sought to determine whether racial disparities exist in ACP among primary care patients. Our experience suggests a difference in end-of-life care expectations and planning between racial groups. We sought to identify the presence of ACPs within existing patients’ EHRs as well as the documentation of advance care discussions. Our objective was to determine whether there were disparities in these outcomes based on recorded race.

Methods

Results

A total of 22,372 patients were included (Table 1). The mean age of the participants was 74 years, 87% were White, 51% were female, 55% were married, and 88.0% were insured by Medicare (Table 1). The total number of patients with combined ACP was 7,678 (34%); those with advanced care discussion documented were 2,298 (10%); and those with documented advanced care directives or healthcare power of attorney were 6,446 (29%).

The estimated effects of race on ACP after controlling for sex, age, marital status, insurance coverage, and updated CCI are provided in Table 2. Black patients and patients of other races were less likely than White patients to have documented ACP (adjusted odds ratio [aOR] = 0.64; 95% confidence interval [CI], 0.58-0.71; P <.0001 and aOR = 0.63; 95% CI, 0.48-0.82; P = .0007, respectively). Older patients and those with a higher updated CCI were more likely to complete ACP (aOR = 1.48; 95% CI, 1.42-1.54; P < .0001 and aOR = 1.32; 95% CI, 1.30-1.35; P < .0001, respectively). Compared to married patients, those who were single, divorced, or widowed were more likely to have a plan (aOR = 1.12; 95% CI, 1.02-1.23, P = .0149; aOR = 1.26, 95% CI, 1.14-1.39, P < .0001; and aOR = 1.35, 95% CI, 1.24-1.46, P < .0001, respectively). Sex and insurance type were not found to be predictors of ACP.

Considering the two components of our composite outcome individually, bivariate associations between patient characteristics and ACP discussion with a clinician are reported in Table 3. Bivariate associations between patient characteristics and the combination of advance directives and healthcare power of attorney are reported in Table 4. A higher proportion of Black patients had an ACP discussion with a clinician compared to White patients and patients of other races (12.8%, 10.0%, and 8.9%, respectively [P < .0001]). Except for ethnicity, all other patient characteristics showed significant bivariate associations. A higher proportion of White patients had advance directives compared to Black patients and patients of other races (30.0%, 21.7%, and 20.3%, respectively [P < .0001]). Except for religion, all other patient characteristics showed significant associations.

Discussion

Our study identified disparities in the ACP composite outcome among Black patients and other non-White patients that exists in a large national healthcare system. These findings suggest that this disparity is true for Black patients alone, as well as other non-White groups when excluding Black patients. The difference found between races in this study is based primarily on the completion and existence of ACP in patient charts, which was proportionally much higher in White patients; Black patients, however, were more likely to have had an ACP discussion documented (Tables 3, 4). While the disparities in ACP identified in this study are consistent with similar findings in other reports, to our knowledge, this is the first study investigating documentation in EHR and CPT billing codes for the existence of ACP. Previous publications have demonstrated disparities at the end of life in multiple ways, including lower rates of completion of Do-Not-Resuscitate (DNR) orders in Black patients compared to Whites during heart failure admissions in California⁶; less ACP in Black and Latino subjects compared to White subjects in the National Health and Aging Trends cohort¹⁶; more frequent completion of advance directives in White subjects of the Healthstyles survey¹⁷; and less likelihood of engaging in ACP among Blacks and Hispanics compared to Whites in the Health and Retirement cohort.⁷ A lack of ACP can increase burdens on both patients and their caregivers. In Black and minority populations in different settings, previous research has reported increased healthcare transitions including trips to the emergency department and inpatient admissions, admissions to the intensive care unit, and both enrollment and disenrollment in hospice care.^12,13,18,19

The cause of disparities in ACP and end-of-life care appears to be multifactorial.²⁰ It is likely that demographic factors like income, education level, and health literacy play a role.^16,17,20 While we did not investigate education, income, or health literacy, our study indicates marital status, increased burden of illness (elevated CCI), and age all play a role in likelihood of ACP. Previous studies have identified widowhood^21,22 and age^23,24 as predictors of ACP completion. Our findings of elevated level of ACP in divorced patients compared to married patients contradicts previously published literature.²¹ Evidence on disease burden is mixed. In some cases, cancer diagnoses and perceived decreased life expectancy can increase the likelihood of ACP,^25,26 while multi-morbidity has been shown to decrease it.²⁵ Our multifactorial analysis controlled for these factors when investigating the outcome of racial disparity on ACP, but did not explicitly investigate the influence of age, marital status, or CCI on the race-based outcome (Table 2).

Trust, or lack of trust, in healthcare systems has also been suggested as a potential cause of racial disparity in ACP.²⁷ Empirical evidence to date has not supported this theory.⁸ While not explicitly stated as trust, there is evidence that Black patients prefer closer relationships to caregivers and expect understanding and respect of cultural beliefs when making end-of-life decisions.¹⁰ Some patients see death as a spiritual issue rather than a medical one. Multiple authors have reported findings relating to religion’s role in end-of-life decision making, which include a desire for information about end-of-life care to come from churches⁹ or a willingness for decisions to be left “in God’s hands.”²⁸ In some cases, spiritual beliefs can influence engagement in end-of-life planning, leading patients to choose less-aggressive interventions.^20,27 These findings may help explain the results included in our bivariate analyses that a higher proportion of Black patients had an ACP discussion with a clinician compared to White patients and patients of other races (Table 3). The numeric disparity in those completing a discussion compared to those with documented ACP in the chart may suggest that the discussion with a medical provider may not necessarily lead to the goal of creating an ACP that is officially documented. Furthermore, in Ohio and other states, ACP can require a notary to make it official, which may pose logistical and financial obstacles to ACP completion.

Similar to the complexity that surrounds reasons for not accepting or completing ACP, the solutions to this problem have also proven difficult and multifactorial. There is evidence that educational interventions for minority groups can improve acceptance and completion.²⁹ However, these interventions require significant time commitments and have often been studied in locations outside the healthcare system. Evidence suggests that education through religious entities and churches is preferred in some parts of the Black community,⁹ as is a focus on direct communication and family-centered discussion and decision-making.²⁰ A qualitative study involving interviews with Black, Latino, and White participants revealed themes centered on the desire for culturally sensitive interactions and communication. Black and Latino participants expressed a preference for interaction with caregivers who share similar characteristics, including race, gender, and religious background.³⁰ There have also been calls for education of caregivers to facilitate more substantive and effective end-of-life discussions.¹

In response to the identified disparities in ACP documentation and completion, our department initiated targeted interventions to better embed ACP into routine primary care workflows. A standard workflow and patient educational packet were developed to both prompt and guide patient discussions about ACP during both routine and annual wellness visits. A standard EHR documentation phrase was created to streamline and allow tracking of these events. Ongoing interdisciplinary meetings are used to monitor implementation progress, identify barriers, and refine the process and documentation. If successful, the next step will be to expand this initiative from one office to a larger primary care setting; if disparities around race and ethnicity remain, we will attempt to identify contributing factors and create steps to address ongoing gaps in care.

Limitations

While this project was strengthened by use of a large searchable database and inclusion of over 22,000 patients, the underrepresentation of non-White patients in comparison to the overall United States population weakens its generalizability. Other limitations are inherent to retrospective chart review studies. For example, our use of the billing code as a proxy for end-of-life discussion makes it likely that we have missed occurrences where discussions occurred but were not billed. It is also probable that not all advance directives or information regarding healthcare power of attorneys are recorded within the EHR, resulting in an underestimation of ACP. Additionally, while we collected demographic data on religion, insurance type, marital status, and CCI, we did not gather information on other potentially important factors such as language, psychological status, education, literacy, or past experiences with death.

Conclusion

Our results show racial disparities in ACP within our population. While Black patients were more likely to have had an ACP discussion, overall, Black and other non-White patients were less likely than White patients to have one of the composite ACP outcomes documented in their charts.