Cardiopulmonary Resuscitation Preferences of People Receiving Dialysis

This cross-sectional survey study describes the cardiopulmonary resuscitation preferences of patients receiving dialysis and how these preferences are associated with their responses to questions about other aspects of end-of-life care.


Introduction
People with end-stage kidney disease (ESKD) receiving maintenance dialysis have mortality rates of 15% to 20% per year. 1 Compared with the general population, these patients are far more likely to die prematurely of a cardiac arrest. [2][3][4][5] Out-of-hospital sudden cardiac death occurs 20 times more often in people receiving dialysis than in the general population. [1][2][3]6 Less than 25% of patients who experience a cardiac arrest during dialysis treatment survive to hospital discharge, and only 8% to 15% survive more than 1 year after the arrest. 3,6-10 Most prior studies have also shown that outcomes after in-hospital cardiac arrest are less favorable for these patients than for more broadly defined populations, with only one-half surviving beyond 5 months (compared with almost 3 years for the general population). 11,12 Patients receiving maintenance dialysis who survive a cardiac arrest in the hospital are also more likely to be discharged to a skilled nursing facility compared with other survivors of cardiac arrest. 13 Despite their generally poor outcomes after resuscitation, available data suggest that patients receiving dialysis are much more likely to receive cardiopulmonary resuscitation (CPR) than members of the general population. These patients are also more likely to receive intensive patterns of end-of-life care focused on life prolongation compared with some other seriously ill populations. 11,[14][15][16][17] The extent to which these aggressive patterns of end-of-life care focused on life extension align with the values, goals, and preferences of individual patients receiving dialysis is not known. 18 Although most patients with advanced kidney disease say that they would want to be resuscitated, [19][20][21] most also indicate that they would value comfort and relief of suffering if they were seriously ill or dying and that they would prefer to die at home. 20 Because resuscitation preferences are often elicited reflexively at times of transition, such as hospital admission or dialysis initiation or when obtaining consent for procedures, these conversations may not be integrated with a broader process of advance care planning that explores patients' values, goals, expectations, and preferences for future care. 22,23 Although at least 1 study 3 has attempted to gauge the extent to which the resuscitation preferences of patients undergoing dialysis are honored, no prior studies to our knowledge have evaluated the association between patients' resuscitation preferences and how they respond to questions about other aspects of end-of-life care. This study used data collected from patients receiving dialysis who participated in a survey about end-of-life care to describe the association of resuscitation preferences with selfreported participant characteristics and with other domains of palliative and end-of-life care.

Study Design
In this cross-sectional survey study, we analyzed data from the United States Renal Data System Study of Treatment Preferences (USTATE) conducted among patients at 31 nonprofit dialysis facilities in 2 US metropolitan areas (Seattle, Washington, and Nashville, Tennessee) between April 22, 2015, and October 2, 2018, in accordance with best practices for survey research. 24,25 Analyses for this article were conducted between December 2018 and April 2020. Patients were eligible to participate in USTATE if they were 21 years or older, sufficiently fluent to complete surveys in English, cognitively able to provide written informed consent, and receiving maintenance dialysis. Study staff consulted with dialysis facility charge nurses to identify patients who met these eligibility criteria and then approached eligible patients in person during their dialysis session, resulting in a pragmatic consecutive sample of eligible patients receiving dialysis at each facility at the time of survey administration. Most patients approached were receiving in-center hemodialysis; a small convenience sample of patients were receiving peritoneal dialysis. Study participants could choose whether to complete the survey themselves or have a study coordinator record their responses to survey questions during their dialysis session. The study was approved by the institutional review board at the University of Washington, Seattle, and written informed consent was obtained from participants. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Survey
The USTATE survey included items that had been previously validated as well as items developed by Respondents were also asked to provide some basic information about themselves (eg, age, sex, race, ethnicity, highest educational level, importance of spiritual and/or religious beliefs, time receiving dialysis, and self-reported health assessment).

Primary Exposure
Responses to the following question were used to define the primary exposure: "If you had to decide right now, would you want CPR (cardiopulmonary resuscitation) if your heart were to stop beating?" Possible responses included definitely yes, probably yes, probably no, and definitely no. For the analyses described herein, participants who indicated they would probably or definitely want CPR were classified as preferring CPR, whereas those who said they would probably or definitely not want CPR were classified as preferring not to be resuscitated (do not resuscitate [DNR]). 32

Outcomes
This study examined the association between preference for CPR and 9 study outcomes. Outcome 1 was the preference for receipt of mechanical ventilation; participants were asked if they would want this intervention if they "had to decide right now," and those indicating they would definitely or probably want this intervention were classified as preferring mechanical ventilation. Outcome 2 was self-reported engagement in advance care planning, which included (1) documentation of a surrogate decision maker ("I have signed official papers naming someone to make medical decisions for me [eg, as part of a living will or advance directive]" and have or have not discussed this choice with him or her) and (2) documentation of treatment preferences ("I have signed official papers documenting my preferences (eg, living will or advance directive)" and have or have not talked with at least one friend or family member about this preference). Outcome 3 was the participant's values around life prolongation. Participants were queried about their preferred plan of care if they "were to become very sick in the future," with responses categorized as "extending life even if that means having more pain and discomfort" vs "relieving pain and discomfort as much as possible, even if that means not living as long" vs "I'm not sure which I would choose". Outcome 4 was whether participants had had thoughts or prior discussions about stopping dialysis vs not. Outcome 5 was whether they had had thoughts or prior discussions about receiving hospice care if they were to become sicker or if their goals changed vs not. Outcome 6 was the respondent's desired place of death (categorized as home if home or the home of a relative or friend or as other if hospital, nursing home, or other setting).
Outcome 7 was the expectation about prognosis (categorized as predicted life expectancy of <5 years vs 5-10 years vs >10 years vs unsure). Outcome 8 was the presence or absence of symptoms over the past week, including (1) weakness or lack of energy, (2) pain, (3) difficulty sleeping, (4) poor mobility, (5) anxiety, (6) shortness of breath, and (7) depression. Outcome 9 was the presence of palliative care needs, including (1) "would like someone to talk to about treatment options for the future" vs not, (2) "would like help with making plans in case I become very ill (advance care planning)" vs not, (3) "would like someone to talk to about my care plan and treatments" vs not, (4) "would like to have someone to talk to about finding meaning in my life now" vs not, (5) "would like help with learning to cope with feelings of sadness" vs not, and (6) "would like to have someone to talk to about dying and death" vs not.

Statistical Analysis
Logistic regression was used to evaluate the association of the aforementioned self-reported participant characteristics with a preference for CPR. To evaluate the association between preference for CPR and the 9 study outcomes, this study used logistic or multinomial regression depending on whether the outcome was binary or multicategorical, after adjustment for self-reported participant characteristics. Estimates of association are presented as adjusted risk differences with 95% CIs.
Predicted risks of the relevant outcome for the exposure group were calculated by fixing the value of the other covariates at the mean value for the cohort. To account for multiple comparisons, a threshold of 2-sided P Յ .001 was used for analyses of study outcomes. All analyses were conducted using Stata, version 13.1 (StataCorp LLC).

Results
Between April 22, 2015, and October 2, 2018, a total of 1592 patients were approached to participate in the study. During the first 8 months of recruitment, 161 eligible patients were invited to participate in the pilot phase of the study, and 146 of these completed pilot versions of the survey (90.7% response rate). After the pilot phase, a further 1431 eligible patients were invited to participate, and 1006 patients completed the final version of the survey. Nine of these respondents were excluded because they did not record their name and/or date of birth on the paper survey or consent form, along with an additional 124 participants who did not complete one or more survey questions included in the analyses described herein, yielding an analytic cohort of 873 participants (61.0% of 1431 eligible patients invited to complete the final version of the survey).

Cohort Characteristics and CPR Preference
Among 873 participants included in the analytic sample, the mean (SD) age at the time of survey completion was 62.6 (14.0) years; 489 (56.0%) were men, and 526 (60.3%) were White individuals.
In the total sample, 248 participants (28.4%) had been receiving dialysis for less than 1 year at the time of survey administration, 426 (48.8%) had been receiving dialysis between 1 and 5 years, and  Table 1).

JAMA Network Open | Nephrology
Among 873 participants, 571 (65.4%) said that they definitely wanted to be resuscitated and 165 (18.9%) said that they probably wanted to be resuscitated. A total of 137 participants did not prefer resuscitation; 60 (6.9%) said that they probably did not want to be resuscitated, and 77 (8.8%) said that they definitely did not want to be resuscitated.

Association Between Self-reported Participant Characteristics and Preference for CPR
Self-reported participant characteristics that were independently associated with a preference for CPR included younger age, Black race, and reporting that spiritual and/or religious beliefs were important to them. These results are summarized in Table 2.

Limitations
Our study has several important limitations. First, the results of this cross-sectional survey study conducted at nonprofit dialysis facilities in 2 US metropolitan areas may not be generalizable to patients receiving dialysis at for-profit facilities, those living in other parts of the country, and those who were excluded from our study (non-English-speaking individuals and persons unable to provide written informed consent). Although the demographic characteristics of our cohort resemble those of the overall US population receiving dialysis, 24 the high percentage of patients enrolled in our study who rated their health as good, very good, or excellent could mean that their CPR preferences are not generalizable to the overall US population undergoing dialysis. Second, although we used an existing instrument to elicit participants' CPR preferences, 32 our approach has some inherent limitations and provides only a partial understanding of these preferences. For example, a Likert-type scale is subject to response bias, 75 and the scale we used did not include neutral or uncertain response options. We also did not frame the question about CPR preference with background information about the intervention and expected outcomes. Nor did we ask about patients' resuscitation preferences in different hypothetical future health states, elicit trade-offs, assess for stability in preferences over time, or collect information on dialysis facility resuscitation policies.
Third, how individuals responded to survey items may differ from responses that they might provide during discussions with clinicians about goals of care because of differences in context.

Conclusions
Most patients undergoing dialysis at 31 nonprofit facilities in 2 US metropolitan areas who participated in this study indicated they definitely or probably wanted to be resuscitated if their heart were to stop beating. Resuscitation preferences were associated with responses to some, but not all, questions about other domains of end-of-life care. Furthermore, participants' CPR preferences were not always aligned with how they responded to questions about these other aspects of end-oflife care. These findings underline the importance of educational initiatives to improve patients' understanding of the clinical context and outcomes of CPR and of contextualizing discussions about code status in a broader conversation and understanding what matters most to each patient.