Visual Abstract
Abstract
Background Patient-centered care for older adults with CKD requires communication about patient’s values, goals of care, and treatment preferences. Eliciting this information requires tools that patients understand and that enable effective communication about their care preferences.
Methods Nephrology clinic patients age ≥60 years with stage 4 or 5 nondialysis-dependent CKD selected one of four responses to the question, “If you had a serious illness, what would be important to you?” Condensed versions of the options were, “Live as long as possible;” “Try treatments, but do not suffer;” “Focus on comfort;” or “Unsure.” Patients also completed a validated health outcome prioritization tool and an instrument determining the acceptability of end-of-life scenarios. Patient responses to the three tools were compared.
Results Of the 382 participants, 35% (n=134) selected “Try treatments, but do not suffer;” 33% (n=126) chose “Focus on comfort;” 20% (n=75) opted for “Live as long as possible;” and 12% (n=47) selected “Unsure.” Answers were associated with patients’ first health outcome priority and acceptability of end-of-life scenarios. One third of patients with a preference to “Focus on comfort” reported that a life on dialysis would not be worth living compared with 5% of those who chose “Live as long as possible” (P<0.001). About 90% of patients agreed to share their preferences with their providers.
Conclusions Older adults with advanced CKD have diverse treatment preferences and want to share them. A single treatment preference question correlated well with longer, validated health preference tools and may provide a point of entry for discussions about patient’s treatment goals.
Older patients with advanced CKD often experience disturbing symptoms, functional decline, long-term institutionalization, or burdensome treatments such as dialysis.1–3 In this setting, nephrologists’ limited understanding of their patients’ values4,5 and missed opportunities to engage their patients in shared decision making may negatively affect patient quality of life, quality of dying, and patient-centered care.6–11 This may also contribute to treatment regret regarding dialysis initiation,6 poorer end-of-life care compared with patients with other serious illnesses,9 and high health care utilization at the end of life.12,13
Leading nephrology organizations have advocated for advance care planning as a clinical priority.14 Unfortunately, nephrologists face multiple barriers with respect to advance care planning, namely, that patients with advanced CKD are medically complex, receive fragmented care, and have uncertain trajectories. Nephrologists also report an unclear locus of responsibility, limited clinical time, and limited communication skills to initiate and lead these discussions.10,15–18 In addition, data on patient treatment preferences and possible advance care planning interventions in patients with CKD are limited.19 Although patients have a variety of life experiences that affect their perspectives toward advance care planning,20 a majority prefer for providers to discuss prognosis and present multiple treatment options when faced with disease progression.6,20,21 However, many providers report reluctance to initiate these discussions, with fewer than 5% of nephrologists reporting a discussion of prognosis with their patients in one study.10,22,23
Efficient measures that can be easily integrated into routine clinical activities may help nephrologists elicit patient values, assess patient readiness to engage in advance care planning conversations, and facilitate more informed, shared decision making discussions regarding intensive treatments, such as dialysis.24 We hypothesized that in older patients with advanced CKD, a single-item instrument that assessed patient preferences when confronted with a hypothetical serious illness would associate with health outcome priorities as well as responses on the acceptability of common, hypothetical end-of-life scenarios. We also hypothesized that most patients would be willing to have this information documented in their medical charts and shared with their providers.
Methods
Study Setting and Participants
Patients followed by a nephrology provider at an academic clinic in the southern United States were enrolled between November 2016 and October 2018. Patient eligibility criteria, as previously described in related studies,5,25,26 were age ≥60 years and nondialysis-dependent CKD stages 4–5, defined by the Modification of Diet in Renal Disease Study equation.27 We excluded patients with clinically apparent moderate to severe dementia, dialysis dependence, history of kidney transplantation, or recent AKI (baseline eGFR ≥60 ml/min per 1.73 m2 within the prior 12 months or baseline eGFR ≥40 ml/min per 1.73 m2 within the prior 4 months).
We approached 484 patients and 405 (84%) consented (Figure 1). For this analysis, 23 patients were excluded (18 who were enrolled before the serious illness question was included in the protocol and five for unrecorded responses to the serious illness question after its inclusion) and the remaining 382 patients are described here. The study was approved by the Vanderbilt University Medical Center Institutional Review Board (approval number 161523) and adhered to the principles of the Declaration of Helsinki.
Study flow diagram.
Serious Illness Treatment Preference Question
As previously described,5 trained research staff collected information regarding health values, including several measures evaluating health outcome priorities. Research staff then asked patients the single-item serious illness treatment preference question adapted from the landmark Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT):28 “If you had a serious illness, what would be important to you?” There were four response options: “I want treatments to try to live as long as possible. I would not want to stop treatment even if I was in pain, could not feed or care for myself, or needed machines to live” (i.e., treatments to live as long as possible); “I want to try treatments for a period of time, but I would not want to suffer. If after a period of time the treatments do not help or I am suffering, I would want to stop.” (i.e., try treatments, but do not suffer); “I want to focus on my quality of life and being comfortable, even if it means having a shorter life.” (i.e., focus on comfort); and “I am not sure.” The interpretation of what constituted a serious illness was left to the patient to define.
Health Outcome Priorities
As previously described, patients also completed a validated health outcome prioritization tool.5,24 Patients were presented with a script and were asked to rank health outcome priorities (“maintaining independence;” “keeping you alive;” “reducing or eliminating pain;” and “reducing or eliminating symptoms” such as dizziness, fatigue, shortness of breath) on an enlarged visual analog scale with a range of 0 (lowest priority) to 100 (highest priority). Patients could not rank two priorities equally. Because the health outcome priority rankings demonstrated excellent reliability in previous studies whereas the numerical visual analog scale scores demonstrated substantial variation, we chose a priori to capture rank orderings only.24
End-of-Life Scenarios
We used a publicly available advance care planning workbook29 to capture patient preferences in common end-of-life scenarios (e.g., cannot recognize family or friends, kept alive by kidney dialysis, and live in a nursing home permanently). Research staff introduced the material, spoke about the importance of understanding the patients’ responses should their illness worsen, and asked patients to carefully consider their values and preferences when answering the items. Each item prompted the patients to judge whether they would find life in a described condition “difficult but acceptable;” “worth living, but just barely;” or “not worth living.” The patients were prompted to consider each of the described scenarios as permanent and were permitted to respond “Can’t answer now” if unable to provide a response or if they found the question distressing. Patients were permitted to opt out of each question and were provided with emotional support if necessary.
Willingness to Share with Providers and Document in Electronic Health Record
Six months after study initiation, we added an additional question immediately after the serious illness item, to assess whether patients would be willing to have their response to the serious illness item shared with their doctors and recorded in their electronic health record (EHR). Response options were yes or no.
Patient Characteristics and Other Measures
Upon enrollment, patients completed a brief questionnaire to collect information on sociodemographics and past medical history. We also performed manual chart reviews and structured data abstraction to supplement this information with documented comorbidities and common clinical measurements (e.g., vital signs, laboratory test values, etc.) from the prior 12 months. With information on comorbidities, we calculated a Charlson comorbidity index (CCI) for each patient.30 We measured independence in activities of daily living (ADL) and instrumental activities of daily living (iADL) with the Katz31 and Lawton32 indices, respectively. ADL and iADL scores ranged from 0 to 5 and 0 to 8, respectively, with higher scores indicating greater independence in activities. We used the Palliative Care Outcome Scale symptom list for end-stage renal disease measure2,33 to assess patient symptoms over the prior week generating a total score, ranging from 0 to 68, with higher scores indicating a higher symptom burden.
Statistical Analyses
Patient characteristics are presented as medians and interquartile ranges (IQRs) for continuous variables and counts with percentages for categorical variables. We tested the association between serious illness treatment preferences and (1) first health outcome priority and (2) acceptability of common end-of-life scenarios with chi-squared tests.
We used binary logistic regression to compare patients who answered “Unsure” to the serious illness treatment preference to the remainder of the cohort. Patients who responded “Unsure” were similar to respondents with a delineated preference. We therefore excluded patients who responded “Unsure” from our multivariable regression model to focus on patients with known preferences. We then modeled serious illness treatment preference responses using multivariable ordinal logistic regression with a rank order starting with aggressive care and progressing to less aggressive care (i.e., “Live as long as possible” ordered as 0; “Try treatments, but do not suffer” ordered as 1; and “Focus on comfort” ordered as 2). Independent variables in the model were selected a priori on the basis of plausibility of an association: age, sex, race, marital status, insurance, income (as a continuous variable in $10,000 U), education, eGFR, Palliative Care Outcome Scale symptom list for end-stage renal disease score, and comorbidities (coronary artery disease, cerebrovascular disease, peripheral vascular disease, heart failure, and diabetes). Variables were checked for collinearity. We repeated the regression analysis using the CCI instead of individual comorbidities. Continuous predictors were included in regression models as restricted cubic splines to allow for possible nonlinear association. For continuous variables, odds ratios (ORs) were scaled to express the odds per an increase equal to the variable’s IQR. For categorical variables, ORs were expressed against specified references.
To better characterize the association between patients’ response to the serious illness treatment preference question and their first health outcome priority, the relationship was assessed using an unadjusted logistic regression model, with the dependent variable being the dichotomized first health outcome priority (staying alive versus any other priority i.e., pain, symptoms, or independence). After a linear association was graphically demonstrated with the serious illness treatment preference response, the serious illness treatment preference response was treated numerically (i.e., “Live as long as possible” ordered as 0; “Try treatments, but do not suffer” ordered as 1; and “Focus on comfort” ordered as 2). In addition, we assessed rank-order correlation of the dichotomized first health outcome priority with the serious illness treatment preference response using the Goodman and Kruskal γ to account for ties.34–36 P values of <0.05 were considered statistically significant. We performed all analyses using R (version 3.4.4).37
Results
The 382 participants included 46% women, 16% black Americans, 62% with at least some college education, and 69% with Medicare insurance (Table 1). Participants had a median age of 71 years (IQR, 66–77), a median CCI score of 6 (IQR, 4–7), a median eGFR at enrollment of 23 ml/min per 1.73 m2 (IQR, 17–28), and most patients were independent in their ADL and iADL (Table 1).
Baseline characteristics by serious illness treatment preference response
Serious Illness Preference
Approximately 33% of patients (n=126) chose “Focus on comfort;” 35% (n=134) chose “Try treatments, but do not suffer;” 20% (n=75) chose “Live as long as possible;” and 12% (n=47) chose “Unsure” as their serious illness treatment preference (P<0.001). Patients who responded “Unsure” were modestly younger and more likely to be black than their cohort counterparts (median age, 68 [IQR, 65–74] versus 71 [IQR, 66–77] years; black, 28% [n=13] versus 14% [n=48]) but were otherwise like respondents with a delineated preference.
Association with Health Outcome Priority and Advance Care Planning Responses
Patients’ serious illness treatment preferences were associated with patients’ first health outcome priority (Figure 2, Supplemental Table 1; P<0.001). Maintaining independence was the first health outcome priority for 55% and 51% of patients who selected “Focus on comfort” or “Try treatments, but do not suffer,” respectively, compared with 34% of patients who selected “Live as long as possible” as their treatment preference (“Live as long as possible” versus “Try treatments, but do not suffer” P=0.02; “Live as long as possible” versus “Focus on comfort” P=0.006; “Try treatments, but do not suffer” versus “Focus on comfort” P=0.55]. In addition, staying alive was the first health outcome priority for 56%, 35%, and 20% of patients who selected “Live as long as possible;” “Try treatments, but do not suffer;” and “Focus on comfort” as their serious illness preference, respectively (“Live as long as possible” versus “Try treatments, but do not suffer” P=0.005; “Live as long as possible” versus “Focus on comfort” P<0.001; “Try treatments, but do not suffer” versus “Focus on comfort” P=0.01).
Patients’ first health outcome priority by serious illness treatment preference. *chi-squared P<0.001.
In the binary logistic regression model, each one-point increase toward less aggressive treatment in the serious illness treatment preference question (e.g., “Live as long as possible” to “Try treatments, but do not suffer”) was associated with an increased odds of a first health outcome priority of pain, symptoms, or maintaining independence (OR, 2.21; 95% confidence interval [95% CI], 1.60 to 3.05) versus staying alive. The rank-order correlation between the serious illness treatment preference and the dichotomized first health outcome priority was moderate (0.46; 95% CI, 0.30 to 0.62). The degree of positive and negative correlation between patients’ serious illness treatment preferences and the first health outcome priority is shown in Supplemental Figure 1.
Patients’ serious illness treatment preference selection was associated with acceptability of the hypothetical end-of-life scenarios (Table 2). Patients who chose “Try treatments, but do not suffer” or “Focus on comfort” as their serious illness preference were more likely to say life was not worth living for all the scenarios except “Cannot walk but can use a wheelchair” (Table 2). Approximately 33% of patients (n=41) who chose “Focus on comfort” reported that a life on dialysis would not be worth living compared with 5% (n=4) of patients who chose “Live as long as possible” (P<0.001; Figure 3).
Serious illness treatment preference and acceptance of common end-of-life scenarios
Serious illness treatment preference and the acceptability of being kept alive by chronic dialysis. *“Focus on comfort” versus “Live as long as possible,” P<0.001; “Focus on comfort” versus “Try treatments, but do not suffer,” P=0.005; “Live as long as possible” versus “Try treatments, but do not suffer,” P=0.01.
Patients’ Willingness to Share Preferences
After an item assessing patients’ willingness to share their care preferences was added to the study protocol, 251 patients (66% of analytic cohort) were asked about their willingness to share treatment preferences with their doctors and have it documented in their EHR. About 88% (n=221 out of 251) of queried patients reported a desire to share their preferences and have them documented. After excluding patients who responded “Unsure” to the serious illness treatment preference item, the results were similar with 91% (n=200 out of 220) of patients reporting a desire to share and document their serious illness care preferences (Supplemental Figure 2, Supplemental Table 2). There was no association between serious illness treatment preference response and willingness to have preferences documented (Supplemental Table 2).
Predictors of Serious Illness Preference Response
In the multivariable model, being a widow/widower (reference: married) (OR, 1.94; 95% CI, 1.02 to 3.68), higher income (OR, 2.32; 95% CI, 1.32 to 4.09), and the presence of cerebrovascular disease (OR, 2.48; 95% CI, 1.30 to 4.73) were associated with a preference toward less aggressive care (i.e., treatment goals that prioritize comfort) (Supplemental Table 3). Not having graduated high school (reference: high school diploma or equivalent) (OR, 0.44; 95% CI, 0.20 to 0.99) and private insurance (reference: Medicare) (OR, 0.54; 95% CI, 0.31 to 0.94) were associated with a preference toward more aggressive care (Supplemental Table 3).
Discussion
We found that older adults with advanced CKD had diverse preferences when confronted with a hypothetical serious illness, and only one fifth reported a preference for treatments to live as long as possible. Further, the single-item serious illness measure captured important information about patient priorities and goals of care. The serious illness response associated with patients’ willingness to accept nearly all common end-of-life scenarios and about one third of patients who preferred treatments that focused on comfort reported that a life on dialysis would be unacceptable. In addition, patients’ serious illness responses associated with their stated primary health outcome priority, assessed using a validated health outcome priority tool.5,24 Importantly, the vast majority of patients were willing to share their serious illness preference with their providers and to have them documented in the EHR.
Across specialties, goals-of-care communication between patients and providers has been shown to improve important patient and health care system outcomes.38–42 In the setting of multiple barriers to advance care planning,10,15–18 a shortage of hospice and palliative care physicians,43 and some patients’ preferences to not involve a palliative care team,6 nephrologists need tools that help to initiate conversations to elicit care preferences. The single-item serious illness treatment preference question is simple, comprehensible to patients, and takes little time to collect. Further, as our results suggest, the measure provides useful insight on patient health outcome priorities and preferences regarding end-of-life treatment choices. These characteristics make it well suited for collection during routine nephrology care and future research targeting improvements in end-of-life care in advanced CKD. In addition, the structured response categories facilitate EHR integration (like the near ubiquitous status of “full code” and “do-not-resuscitate” orders).
Generalizable strategies that improve patient–provider communication about treatment goals are needed in nephrology. Although lengthy educational material and preappointments with social workers have been shown to have a moderate effect on advance care planning discussions and documentation in the general population and cancer patients,44–47 many of the tools currently used lack evidence48 or are not practical for routine use during a nephrology office visit because of their length or resource burden. Recently, a preconversation, communication-priming questionnaire during routine outpatient visits in general patients with a serious illness was shown to increase occurrence, documentation, and quality of goals-of-care communication.49 Our findings indicate that the serious illness treatment preference question provides general insight on patient care preferences and should be studied further as an easy-to-use tool to (1) assess patient readiness for advance care planning, (2) prime patients for more in-depth advance care planning conversations, and (3) give providers an entryway to discuss these issues and improve patient–provider communication about goals of care.50 Notably, nearly 90% of patients were agreeable to documenting their serious illness preference response, indicating the measure is likely to be acceptable and pragmatic to integrate into routine medical care. Future studies should examine the stability of serious illness treatment preferences in patients with advanced CKD over time and the potential effect of available responses on surrogate decision makers during severe illnesses.
Nephrologists report difficulty in identifying patients most appropriate for conservative management or a time-limited dialysis trial.51 Patients with advanced CKD who choose to “Focus on comfort” or to “Try treatments, but do not suffer” are likely to benefit from a more thorough discussion of these options. Such patients were more likely to select “Maintaining independence” as their first health outcome priority and a majority reported that life was not worth living if they required full-time care or lived in a nursing home; outcomes that are common in morbid or frail older adults who initiate chronic dialysis.3,52 Nearly one third of patients who preferred treatments that “Focus on comfort” reported that a life on dialysis would be unacceptable, possibly helping nephrologists identify patients who are more likely to perceive conservative care as consistent with their values and preferences.6 In addition, this approach may mitigate the continual pressure to reconsider dialysis in patients with advanced CKD with a clearly stated preference for conservative care.8 For example, EHR documentation of a stable patient preference for comfort-focused care may give providers adequate reassurance on the consistency of patients’ choices concerning dialysis.
Strengths and Weaknesses
Our study has several strengths. First, older adults with advanced CKD frequently receive high-intensity care near the end of life, yet providers have a limited understanding of their preferences.5 Our results give voice to the diversity of these preferences. Second, the number of participants in our study was relatively high compared with relatable studies in advanced CKD. Third, our participation rate was also high reducing selection biases. Fourth, the serious illness treatment preference measure we used was simple, thereby enhancing generalizability and facilitating future adoption.
Our study also has several limitations. First, we measured patient responses to a hypothetical serious illness, which may not reflect patient preferences when faced with the actual situation. Patient treatment preferences are affected by perceptions of treatment burden, possible outcomes, and the likelihoods of these outcomes53; factors we cannot account for in the absence of a firm definition of “serious illness.” However, the serious illness question is meant to catalyze shared decision making and advance care planning, not replace it, and may have wider clinical adoption than lengthier questionnaires.54 Second, we only measured patient responses once. Future studies should examine whether serious illness treatment responses map with actual end-of-life decisions. Third, our patient population was largely white and had high educational attainment, possibly limiting generalizability of our results. Fourth, although patients with clinically diagnosed moderate to severe dementia were excluded, patients with other levels of cognitive impairment were included. Hence, some patients’ subclinical cognitive impairment may have influenced their ability to communicate their true preferences. Fifth, the correlation between the serious illness treatment preference and the first health outcome priority was only moderate, indicating that this tool should not act as a substitute for more meaningful, in-depth advance care planning discussions. Future research should examine whether completion of the serious illness treatment preference question affects initiation of advance care planning conversations or health outcome priority concordance with providers. Finally, we did not capture the treatment preferences of patients’ health care proxies. These individuals are often tasked with medical decision making near the end of life.
In conclusion, our findings show that more than two thirds of older adults with advanced CKD would prioritize treatments that focus on comfort or do not cause suffering when faced with a potential serious illness. Almost all patients were willing to share these preferences with their providers and to document them in their record. Patients’ serious illness treatment preferences also captured important differences in the acceptability of aggressive therapies, including dialysis. The simple, single-item serious illness treatment measure may serve as an important point of entry for initiating more frequent and effective goals-of-care communication.
Disclosures
Dr. Abdel-Kader reports grants from Satellite Health Care Dr. Siew reports personal fees from Akebia Therapeutics Inc., other from DaVita, outside the submitted work; and received honorarium as a contributing author for UpToDate, as well as serving as Associate Editor for CJASN.
Funding
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases grant K23DK090304 (to Abdel-Kader) and National Center for Advancing Translational Sciences Clinical and Translational Science Awards UL1TR000445 and UL1TR002243, all from the National Institutes of Health, the Satellite Health Norman Coplon Extramural Grant Program (Abdel-Kader), and the Vanderbilt Center for Kidney Disease.
Supplemental Material
This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2019040385/-/DCSupplemental.
Supplemental Table 1. Serious illness treatment preference and first choice health outcome priority.
Supplemental Table 2. Patient willingness to share serious illness preference with provider and medical chart.
Supplemental Table 3. Odds of a serious illness treatment preference choice favoring less aggressive care.
Supplemental Figure 1. Visual assessment of association between patients’ serious illness treatment preference and first health outcome priority.
Supplemental Figure 2. Patients’ willingness to share serious illness treatment preference.
Acknowledgments
Abdel-Kader designed the study. Baddour, Salat, and Abdel-Kader collected the data. Mason and Stewart analyzed the data. All authors participated in data interpretation. Baddour and Abdel-Kader drafted the manuscript. All authors revised the manuscript and approved the final version of the manuscript. Each author contributed important intellectual content during manuscript drafting or revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved.
We thank Dr. Beatrice Concepcion for guidance and assistance in creating the visual abstract. We also thank the patients and providers at the Vanderbilt University Medical Center nephrology clinic for their participation during this study.
The funders of this study had no role in study design, data collection, analysis, interpretation, manuscript preparation, or the decision to report the findings.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
- Copyright © 2019 by the American Society of Nephrology
References
In this issue
