Prevalence and Contents of Advance Directives of Patients with ESRD Receiving Dialysis

Abstract

Background and objectives

ESRD requiring dialysis is associated with increased morbidity and mortality rates, including increased rates of cognitive impairment, compared with the general population. About one quarter of patients receiving dialysis choose to discontinue dialysis at the end of life. Advance directives are intended to give providers and surrogates instruction on managing medical decision making, including end of life situations. The prevalence of advance directives is low among patients receiving dialysis. Little is known about the contents of advance directives among these patients with advance directives.

Design, setting, participants, & measurements

We retrospectively reviewed the medical records of all patients receiving maintenance in–center hemodialysis at a tertiary academic medical center between January 1, 2007 and January 1, 2012. We collected demographic data, the prevalence of advance directives, and a content analysis of these advance directives. We specifically examined the advance directives for instructions on management of interventions at end of life, including dialysis.

Results

Among 808 patients (mean age of 68.6 years old; men =61.2%), 49% had advance directives, of which only 10.6% mentioned dialysis and only 3% specifically addressed dialysis management at end of life. Patients who had advance directives were more likely to be older (74.5 versus 65.4 years old; P<0.001) and have died during the study period (64.4% versus 46.6%; P<0.001) than patients who did not have advance directives. Notably, for patients receiving dialysis who had advance directives, more of the advance directives addressed cardiopulmonary resuscitation (44.2%), mechanical ventilation (37.1%), artificial nutrition and hydration (34.3%), and pain management (43.4%) than dialysis (10.6%).

Conclusions

Although one-half of the patients receiving dialysis in our study had advance directives, end of life management of dialysis was rarely addressed. Future research should focus on improving discernment and documentation of end of life values, goals, and preferences, such as dialysis–specific advance directives, among these patients.

Introduction

Dialysis prolongs life for many patients with ESRD. Patients receiving dialysis have higher morbidity and mortality rates than age-matched peers (1). About one quarter of patients receiving dialysis withdraw dialysis at the end of life (EOL) stage (1). Cognitive impairment in patients receiving dialysis is threefold greater than in age-matched peers (2). Patients receiving dialysis are more likely to receive high-intensity care at the EOL, including hospitalization, treatment in the intensive care unit, and life-sustaining treatments—practices that are not in line with what patients say that they want (3). Furthermore, one half of patients lack decision-making capacity at the time that dialysis is withdrawn; thus, the complicated decision of whether to withdraw dialysis is often the responsibility of surrogates (4). Such EOL decision making can be distressing for surrogates when patients’ wishes are unknown.

Advanced care planning (ACP) is a process of communication among a patient, the patient’s loved ones, and health care providers during which the patient’s values, preferences, and goals are identified and from which the rationale that guides future health care decision making is derived. Advance directives (ADs) are legal documents that allow patients to record their wishes for future health care, including EOL care. In the general population, ADs have been associated with decreased health care costs, decreased in–hospital deaths, and increased use of hospice (5). ACP, which often includes the completion of an AD, has been shown to improve surrogate bereavement (6) and increase hospice utilization in the dialysis population specifically (7). Prior studies have shown low prevalence of ADs in dialysis populations (8,9). The Renal Physicians Association recommends that patients receiving dialysis undergo ACP and execute ADs (10). Planning should occur before initiation of dialysis and be iterative over time as clinical circumstances evolve. Patients receiving dialysis should address dialysis management in their ADs, providing clear information on preferences for dialysis management at the EOL stage. Such communication allows respect for the patients’ values, while minimizing the burdens of surrogate decision making.

However, among patients receiving dialysis who have ADs, little is known about the contents of their ADs, such as type of AD (e.g., power of attorney for health care [POAHC], living will, or physician order for life-sustaining treatment [POLST]) and whether their ADs address management of dialysis at the EOL specifically. The objective of this study was to retrospectively determine the prevalence and contents of ADs among patients receiving maintenance in–center hemodialysis through a single health care center.

Materials and Methods

The Mayo Clinic Institutional Review Board approved this study. In accordance with Minnesota law, only patients with written authorization to use their electronic health records (EHRs) for research were included in this study. No stipend was provided.

The Division of Nephrology and Hypertension at Mayo Clinic’s campus in Rochester, Minnesota maintains a database of all patients who undergo dialysis of any kind through our institution. Using this database, we generated a list of patients who received maintenance in–center hemodialysis from January 1, 2007 to January 1, 2012 in the two dialysis units located in Rochester, Minnesota. We chose this timeframe to avoid crosscontamination of data with a quality improvement project involving ACP that started in 2012 within one of the dialysis units. These dialysis units are not for profit and included one free–standing outpatient dialysis unit and one hospital–based outpatient dialysis unit. We excluded the patients who lacked authorization on file to use their EHRs for research purposes. We hand abstracted the demographic and clinical data of the identified patients using the EHR at Mayo Clinic in Rochester by searching demographic information and provider documentation (e.g., clinical notes, admission history and physical examinations, or discharge summaries). Clinical diagnoses were only accepted if they were documented in provider notes. Patients were recorded as deceased if they had died before January 1, 2012. For live patients, age was recorded as of the end of the study period (January 1, 2012). For deceased patients, age was recorded at the time of death. Palliative care consultation was hand abstracted from the EHR if there was a subspecialty palliative care consultation note (either inpatient or outpatient) between January 1, 2007 and January 1, 2012.

Hand abstraction of data from the EHR also included presence of an AD in the record and the AD’s content. We categorized ADs as POAHC if the AD only documented a surrogate decision maker and as living will if the document only addressed care wishes. We categorized the document as combined if the document included elements of both a POAHC and a living will. The category of other was used for physician order forms documenting ADs, such as POLST forms (introduced in Minnesota on March 30, 2010; http://www.mnmed.org/getattachment/advocacy/improving-health-of-minnesotans/POLST-Communications/POLSTform.pdf.aspx?lang=en-US) (11) or nursing home–specific resuscitation forms. Content of ADs was examined for EOL care wishes, any mention of dialysis, and clear instructions on how to manage dialysis at the EOL stage. At each new outpatient visit or hospitalization, patients are asked to submit a copy of their ADs for incorporation into the EHR; if they have no AD, they are offered the forms and assistance in completing one. Although Mayo Clinic has its own AD (http://mayoclinichealthsystem.org/locations/mankato/for-patients/advance-care-planning/create-your-advance-directive) (12), which is a combined form, it accepts and includes in the EHR any AD document that the patient provides. ADs of any kind are scanned and electronically filed into the patient’s EHR. If multiple ADs were present, only the most recent AD—the legally active AD—was used for this study.

Overall group data were summarized, and data were stratified into two subgroups according to the presence or absence of ADs in the medical record. Subgroup data were also summarized. To test for an association with having an AD in the medical record, the subgroups were compared using the two–sample t test for continuous variables and the Fisher exact test or chi-squared test for categorical variables. JMP statistical software, version 8 (SAS Institute Inc., Cary, NC) was used to perform all analyses. A P value <0.05 was deemed statistically significant.

Results

From January 1, 2007 to January 1, 2012, a total of 841 patients received outpatient in–center maintenance hemodialysis through dialysis units at Mayo Clinic in Rochester. Thirty-three patients did not provide authorization to use their EHRs and were excluded, leaving 808 patients eligible for the study. Demographic data are summarized in Table 1. Mean (SD) age of patients receiving dialysis was 68.6 (17.2) years old. A majority of the patients were men (61.2%), and 396 (49.0%) had ADs. Patients who had ADs were older (74.5 versus 65.4 years old; P<0.001), more likely to be white (93.1% versus 80.1%; P<0.001), and more likely to be deceased (64.4% versus 46.6%; P<0.001) than patients who did not have ADs. More patients with ADs received a palliative care consultation than patients without ADs (20.2% versus 13.3%).

Table 1.

Demographic characteristics of the 808 study patients

Characteristic	Patients			P Value
	All, N=808	Without AD, n=412	With AD, n=396
Age, yr, mean (SD)	68.6 (17.2)	65.4 (18.6)	74.5 (14.7)	<0.001
Men, no. (%)	495 (61.2)	283 (68.7)	219 (55.3)	<0.001
Race/ethnicity, no. (%)
White	699 (86.5)	330 (80.1)	369 (93.1)	<0.001
Other than white	109 (13.5)	82 (19.9)	27 (6.6)	<0.001
Residence, no. (%)
Minnesota	595 (73.6)	281 (68.2)	314 (79.3)	<0.001
Other	213 (26.4)	131 (31.8)	82 (20.7)	<0.001
Mortality status, no. (%)
Deceased	447 (55.3)	192 (46.6)	255 (64.4)	<0.001
Palliative care consultation	135 (16.7)	55 (13.3)	80 (20.2)	0.01

AD, advanced directive.

Table 2 summarizes clinical data. Patients with ADs were more likely to have dementia (13.4% versus 7.3%; P<0.01), coronary artery disease (55.6% versus 45.1%; P=0.003), congestive heart failure (57.6% versus 22.8%; P<0.001), and stroke (17.2% versus 10.4%; P<0.01) than patients who did not have ADs.

Table 2.

Clinical characteristics of the 808 study patients

Characteristic	Patients, No. (%)		P Value
	Without AD, n=412	With AD, n=396
COPD	86 (20.9)	102 (25.8)	0.11
Dementia	30 (7.3)	53 (13.4)	<0.01
DM	206 (50.0)	209 (52.8)	0.44
CAD	186 (45.1)	220 (55.6)	0.003
CHF	94 (22.8)	228 (57.6)	0.001
Stroke	43 (10.4)	68 (17.2)	<0.01
Cancer	93 (22.6)	82 (20.7)	0.55

AD, advanced directive; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; CAD, coronary artery disease; CHF, congestive heart failure.

The types and contents of the ADs are summarized in Table 3; 12 of 59 documents categorized as other were POLST documents, 38 were home care– or nursing home–specific documents delineating patients’ resuscitation wishes that predated the introduction of POLST in Minnesota, two were guardianship papers identifying the legal decision maker, and seven were anatomic bequests that had nothing to do with medical decision making. Of the ADs, 123 (31%) were executed before initiation of dialysis. Dialysis (10.6%) was mentioned less frequently than other EOL interventions, including cardiopulmonary resuscitation (44.2%), mechanical ventilation (37.1%), artificial nutrition and hydration (34.3%), and pain management (43.4%). Of the 42 patient ADs that mentioned dialysis, only 13 addressed how to manage dialysis at the EOL stage; of these, 11 provided instructions on discontinuation of dialysis, and two expressed wishes to continue dialysis in all circumstances. Table 4 provides examples of documentation regarding dialysis found in the ADs of these patients receiving dialysis.

Table 3.

Type of advanced directive and life-sustaining treatments addressed in the 396 advanced directives

Characteristic	ADs, No. (%)
Type of AD
Combined (living will and power of attorney)	274 (69.2)
Living will	31 (7.8)
POAHC	32 (8.1)
Other	59 (14.9)
POLST form	12 (3.0)
NH/home care resuscitation forms	38 (9.6)
Guardianship papers	2 (0.5)
Anatomic bequests	7 (1.8)
Life-sustaining treatment
General	316 (79.8)
Artificial nutrition or hydration	136 (34.3)
CPR	175 (44.2)
Mechanical ventilation	147 (37.1)
Dialysis	42 (10.6)
Dialysis management at EOL stage	13 (3.0)
Pacemaker	8 (2.0)
EOL treatment wishes
Pain control	172 (43.4)
Comfort measures	176 (44.4)

AD, advanced directive; POAHC, power of attorney for health care; POLST, physician order for life-sustaining treatment; NH, nursing home; CPR, cardiopulmonary resuscitation; EOL, end of life.

Table 4.

Examples of dialysis documentation in advance directives of patients receiving dialysis

“I do not want dialysis.”

“Use dialysis as appropriate.”

“Continue dialysis as needed.”

“No dialysis.”

“To dialyze or not will be my choice.”

“Do not start dialysis.”

“I have a dialysis catheter in my right arm. No needles or blood pressure cuff in the right arm.”

“If I am permanently unconscious, I would want to stop dialysis.”

“If I am dying, I would want to stop all life–sustaining measures, including mechanical ventilation, artificial nutrition, and dialysis.”

Discussion

Our study, the largest of its kind, determined the prevalence and contents of ADs among patients with ESRD receiving dialysis. About one half (49.0%) of our patients with ESRD receiving dialysis had ADs in their EHRs. White patients were more likely to have ADs than nonwhite patients. These results are similar to other studies involving the rates of ADs in the population with renal failure (8,13–15). Our patients with ADs were older, had more comorbidity, and were more likely to be deceased at the time of data abstraction than patients without ADs. Several explanations are possible for these findings. The finding that patients with ADs were more likely to be deceased than patients without ADs may be because the patients who have ADs were older and sicker and had poorer prognoses. Such persons may be predisposed to execute ADs because they are more cognizant of their limited life expectancy. Furthermore, health care providers may advocate more tenaciously for ADs in such patients. Sicker patients with more complex comorbidities may interact more frequently with various health providers, offering more opportunities for ACP and AD advocacy from multiple sources.

In addition, patients with ADs were more likely to have had a palliative medicine consultation than those without ADs—an intriguing finding, because one might postulate that patients without ADs would have less clear goals of care than those with ADs and, therefore, were more in need of goal clarification. Alternately, patients with ADs may have been sicker (as evidenced by their increased mortality and morbidity rates), leading to more palliative medicine consultation. Palliative medicine consultation may have led to increased completion of ADs, or the presence of an AD triggered a more careful assessment of goals of care, resulting in a palliative medicine consultation. Additional investigation would be necessary to sort these factors out.

Of the study patients who had ADs, only 10.6% of the ADs specifically mentioned dialysis. More detailed analysis results in an even bleaker story. Of the 42 ADs that mentioned dialysis, only 13 addressed how to manage dialysis at the EOL stage. The other 29 ADs were not helpful in informing future dialysis management, although they technically mentioned dialysis. In fact, many of these 29 ADs actually further confused the issue with contradictory statements, such as “I don’t want to start dialysis” for a patient already receiving dialysis. These findings are unfortunate, because patients with ESRD interact frequently with health care professionals and are hospitalized often. The findings also highlight the need for health care professionals to use the numerous encounters that they have with these patients as opportunities to review and discuss their patients’ ACP and ADs.

Although only 10.6% of ADs in our study specifically addressed dialysis, many ADs addressed other life–prolonging or EOL treatments, such as mechanical ventilation, artificial nutrition, and pain control. These results are similar to those of other studies involving patients with heart failure (16), left ventricular assist devices (17), implantable cardiac defibrillators (18), and pacemakers (19). Today, many AD forms prompt patients to document their EOL preferences regarding specific life–prolonging treatments. For example, the Minnesota POLST form specifically prompts patients to address artificial nutrition and hydration, cardiopulmonary resuscitation, mechanical ventilation, and antibiotics (http://www.mnmed.org/getattachment/advocacy/improving-health-of-minnesotans/POLST-Communications/POLSTform.pdf.aspx?lang=en-US) (10). Our Mayo Clinic AD specifically addresses cardiopulmonary resuscitation and pain management at the EOL stage (http://mayoclinichealthsystem.org/locations/mankato/for-patients/advance-care-planning/create-your-advance-directive) (11). It is likely that the contents of the AD in our study correlate principally with specific prompts that happened to be detailed in the specific AD form used. Simply adding hemodialysis as one of the listed treatments in the form might increase the number of ADs documenting patients’ preferences for initiating hemodialysis, but these types of prompts can create confusion for patients already receiving dialysis, which was evidenced in some examples presented in Table 4. These examples are quotes taken from actual ADs of patients already receiving dialysis. The opportunities for misunderstanding abound when a patient receiving dialysis states “no dialysis” in the AD. For patients who are already receiving dialysis, decisions about withdrawal of dialysis are far more pertinent. It seems unrealistic to expect generic, universal AD forms to include specific details surrounding the circumstances of dialysis withdrawal.

Several studies in the 1990s evaluated patient acceptability for disease-specific ADs with mixed results (20–22). We are not aware of any research examining the effectiveness of such disease–specific ADs for patients with ESRD receiving dialysis. Given patients’ willingness to contemplate and document EOL preferences regarding specific life–prolonging treatments with prompts within generic AD forms, one can speculate that disease-specific documents in the form of either disease-specific ADs for those without an AD or perhaps, a disease-specific addendum to existing ADs that specifically address patients’ preferences for dialysis management at the EOL stage would be worth additional investigation.

Finally, patients receiving in-center hemodialysis frequently interface with health care systems and professionals (e.g., multiple times per week at dialysis centers). These frequent encounters among a high-risk population represent a unique opportunity for ACP. Institutions should formulate and implement programs of ACP for such patients. As evidenced by our results (one half of patients had ADs, but only 3% of the ADs addressed dialysis management at the EOL stage), the simple presence of an AD is not the ultimate goal. The goal is to improve patient care by providing care that is congruent with individual values and goals. ADs may improve care by serving as both communication and documentation tools around which to structure conversations surrounding patient values and goals relevant to individual underlying health conditions, including ESRD. The effectiveness of disease-specific ADs over traditional generic AD forms in clarifying patients’ ACP warrants additional investigation.

Our study has limitations. The patient population was limited to the dialysis practice of a single health care institution and may not be generalizable to other practices. We did not collect demographic data on socioeconomic status, marital status, educational background, or payer status that may have further informed us regarding additional barriers to AD completion. Our study population, although accurately reflecting our geographic region, is predominantly white and may not be generalizable to other ethnicities. Patients may have executed ADs but did not provide a copy for inclusion in the EHR, resulting in underestimation of the prevalence of ADs for this population. We believe that the number of such patients is small given our institution’s systematic approach to repeatedly assess ADs at multiple visit opportunities. Finally, our study is retrospective; our institution may be more efficient at facilitating AD execution in 2016.

Our study has several strengths. Unlike prior studies that have been survey based, our study evaluated the data associated with affected patients—their EHRs and actual AD documents. To our knowledge, this is the largest population of patients receiving dialysis in which prevalence of ADs has been determined and the only study to assess the contents of actual ADs.

Although one half of patients receiving dialysis in this study had ADs, only 3% of the ADs actually addressed how to manage dialysis at the EOL stage, an issue of critical importance to patients, physicians, and surrogate decision makers. Future research should not be directed at measuring the simple presence or absence of an AD for patients receiving dialysis. Rather, future investigations should aim to identify strategies that improve discernment and documentation of EOL values, goals, and preferences among these patients. Exploring dialysis-specific ADs is one potential future investigational strategy.

Disclosures

P.S.M. is a member of the Boston Scientific Patient Safety Advisory Board and an associate editor for New England Journal of Medicine Journal Watch General Medicine. The other authors have no conflicts of interest or financial disclosures to report.