Article Text
PDF
Abstract
Objectives To quantify agreement between patients and their family members on their own values and preferences for use or non-use of life-sustaining treatments for the patient.
Methods Hospitalised patients aged 55 years or older with advanced pulmonary, cardiac, liver disease or metastatic cancer or aged 80 years or older from medical wards at 16 Canadian hospitals and their family members completed a questionnaire including eight items about values related to life-sustaining treatment and a question about preferences for life-sustaining treatments.
Results We recruited a total of 313 patient-family member dyads. Crude agreement between patients and family members about values related to life-sustaining treatment was 42% across all eight items but varied widely: 20% when asking how important it was for the patient to respect the wishes of family members regarding their care; 72% when asking how important it was for the patient to be kept comfortable and suffer as little as possible. Crude agreement on preferences for life-sustaining treatment was 91% (kappa 0.60; 95%CI 0.45 to 0.75) when looking at preferences for cardiopulmonary resuscitation (CPR) versus no CPR but fell to 56% when including all five response options with varying degrees of resuscitative, medical or comfort options (kappa 0.39; 95%CI 0.31 to 0.47).
Conclusions There is appreciable disagreement between seriously ill hospitalised patients and family members in their values and preferences for life-sustaining treatment. Strategies are needed to improve the quality of advance care planning, so that surrogates are better able to honour patient’s wishes at the end of life.
- advance care planning
- goals of care
- life-sustaining treatment
- values
- patient-family member dyads
- agreement
Statistics from Altmetric.com
- advance care planning
- goals of care
- life-sustaining treatment
- values
- patient-family member dyads
- agreement
Background
Almost half of hospitalised older adults will lose decision-making capacity and depend on family members to make decisions about treatment on their behalf.1 These patients tend to have complex medical stays and higher inhospital mortality, so family members are often put in a position of making important decisions about life-sustaining treatment on behalf of their loved ones.2 3 Surrogate decision making must uphold the standard of substituted judgement, whereby surrogates makes decisions that are consistent with patients’ prior expressed wishes. However, if advance care planning activities do not occur or are of low quality, then family members may not be aware of a loved one’s values and preferences for life-sustaining treatment. Previous studies suggest that sometimes family members ‘project’ their own desires for their loved ones rather than reflecting patients’ wishes.4 5 Surrogate decision making can thus be compromised when (1) a family member is unaware of the patient’s values or (2) a family member’s values and preferences are different from those of the patient, and the family member allows these differences to influence decisions they make on behalf of their loved one.
Previous studies have found that family members incorrectly predict patients’ end-of-life (EOL) treatment preferences one-third of the time.6–8 However, relatively few studies have examined family members’ own values and preferences for life-sustaining treatment, and few have focused on current preferences for care for seriously ill patients and their family members.4 9 We conducted this study to investigate the level of agreement between (1) seriously ill hospitalised patients’ values and preferences for life-sustaining treatment and (2) their family members’ own values and preferences for their loved ones’ life-sustaining treatment (ie, we did not ask family members to predict what they thought the patient would want). A secondary objective was to explore whether demographic variables, the type of patient–family member relationship, previous experiences making EOL decisions for others and previous engagement in advance care planning could predict the extent of agreement between patients and their family members regarding preferences for use or non-use of life-sustaining treatments.
Methods
Study population
Between September 2011 and March 2015, the ACCEPT (Audit of Communication, CarE Planning and DocumenTation) study group conducted a multicentre prospective study, including three sequential audits of advance care planning at 16 acute care hospitals across Canada. The details of study methods have been published elsewhere.10 Hospitalised patients with advanced pulmonary, cardiac, liver disease or metastatic cancer or any patients aged 80 years or older who were admitted for an acute medical condition were invited to participate in the study (see online supplementary appendix A for detailed patient eligibility criteria). Patient participants were asked to identify an adult family member who they felt knew them best (including close friends) and who had visited them at least once during the current hospitalisation (excluding paid caregivers). These family members were also invited to participate in the study. For this paper that examines patient–family member agreement, we only include dyads (ie, patient–family member pairs) in which both the patient and family member participated.
Supplementary Material
Supplementary Appendix 1Data collection
Questionnaires were separately administered face to face to seriously ill hospitalised patients and family members to assess demographic information, engagement in advance care planning activities, previous experience in making EOL medical decisions for another person and, for family members, the nature of their relationship to the patient. A global measure of frailty that was adapted from the Clinical Frailty Scale was also included in the questionnaire.11 The development and validation of the questionnaire used in this study have been previously described.12
Eliciting participant values about life-sustaining treatment
The study questionnaire includes eight items that asked participants to rate the importance they place on specific values related to the use of life-sustaining treatment (table 1). These items were rated on a 10-point scale (1=‘not at all important’ and 10=‘very important’) or ‘unsure or undecided.’ We asked family members to indicate their own values, rather than predicting the patient’s values. The development of these items regarding values about the use or non-use of life-sustaining treatments for seriously ill, hospitalised patients was based on (1) previous work done by Parks and Winters on the measurement of EOL values among community-dwelling elderly individuals13 14; (2) judgements about clinical sensibility from 25 investigators in the field of EOL communication and decision making with expertise in critical care medicine, palliative care medicine, nursing, research methodology and psychometrics who were in attendance at a biannual research protocol development meeting of the Canadian Researchers at the End of Life Network; and (3) pilot testing of a draft version of these questions with 27 patients and 24 family members at two hospital sites in Canada.12
Agreement between patients’ and family members’ own values related to life-sustaining treatment
Eliciting participant preferences for use or non-use of life-sustaining treatment
The questionnaire asked participants to identify the medical treatment option they would select at this time if life support was needed to keep the patient alive. In the questionnaire used in the third audit cycle, patient and family participants were asked to select from one of six possible medical treatment options: (1) ‘use machines and all possible measures including CPR with a focus on keeping me/them alive at all costs’; (2) ‘use machines and all possible measures with a focus on keeping me/them alive but if my heart stops, no resuscitation’; (3) ‘use machines only in the short term to see if I/they will get better but if my illness is prolonged, change focus to comfort measures only. If my/their heart stops, no resuscitation (CPR)’; (4) ‘use full medical care to prolong my/their life but if my heart of breathing stops, no resuscitation (CPR) or breathing machines’; (5) ‘use comfort measures only, with a focus on improving quality of life and comfort, allow natural death and no artificial prolongation of life and no resuscitation’; or (6) ‘unsure’. In the first two audit cycles, the questionnaire had included an additional option that contained a ‘mix’ of the above options but without an explicit statement about the inclusion of CPR or breathing machines. Consistent with previous analyses, due to perceived ambiguity, this ‘mix’ option was not included in the analyses for this paper.15 As with the questions about values, when asking about life support, we asked family members to indicate their own medical treatment preferences for the patient, not what they thought the patient would select.
Statistical methods
Descriptive statistics were used to present patient and family demographics, as well as patient and family members’ reported prevalence of engagement in specific advance care planning activities and experience in making EOL decisions for someone else. In addition, we used descriptive statistics to report patients’ and family members’ values and preferences for life-sustaining treatment. We determined means and SD for continuous variables and proportions for categorical variables.
To measure patient–family agreement in values, we calculated crude agreement for each of the eight items about values related to life-sustaining treatment. Because examination of exact agreement on a 10-point scale may be overly stringent and underestimate alignment in values within dyads, we performed a sensitivity analysis in which we measured approximate agreement on the 10-point rating scale by calculating the proportion of patient–family dyads that differed by a numeric rating of 2 or less in either direction. In these analyses, we excluded ‘unsure’ and missing response options.
To measure patient–family agreement in preferences for life-sustaining treatment, we first assessed crude agreement, both using all five response options as well as with the response options dichotomised into ‘CPR’ (response option 1) and ‘no CPR’ (response options 2–5). To measure chance-corrected agreement, we calculated the kappa coefficient and corresponding 95% CI. In all of the analyses of preferences for life-sustaining treatment, we excluded cases where either patient or family member participants responded ‘unsure’ or where there was a missing response.
Finally, we examined whether agreement between patients and family members about preferences for life-sustaining treatment was predicted by demographic variables, engagement in specific advance care planning activities (ie, completion of advanced directive, previous discussion about prognosis with physician, assigned surrogate decision maker) and previous experience in making EOL decisions for someone else (see table 2 for all variables included in the model). We selected these variables for inclusion in a multivariate model because we hypothesised a priori that they could be determinants of agreement. We then used backward stepwise logistic regression modelling to identify the strongest independent predictors of agreement. The dependent variable was a binary variable—patient–family agreement or disagreement on the five medical treatment preference options. We excluded the ‘unsure’ and ‘missing’ responses from this analysis.
Predictors of patient–family agreement on preferences for life-sustaining treatment
All statistical analysis was conducted using SAS V.9.3 (SAS Institute, Cary, North Carolina, USA) software. This study was approved by the research ethics board of each participating institution.
Results
During the three audit cycles of our study, we enrolled 816 (56%) of 1466 eligible patients and 642 (67%) of 962 eligible family members. Of these participants, this paper is restricted to analysis of the 313 cases (dyads) in which both the patient and family member took part in the study (figure 1). Most patients (76%) had previously discussed preferences for life-sustaining treatments with someone, and a similar proportion of family members (77%) reported having discussed their relative’s preferences for life-sustaining treatments. Demographic data and descriptive data about past experiences with advance care planning activities are presented in table 3 for both patient and family member participants.
Participant demographics and previous involvement in advance care planning-related activities
Recruitment flow diagram.
Agreement between patients’ and family members’ own values related to life-sustaining treatment
Mean scores (SD) for patient and family responses for each of the eight items about values are shown in table 1. Of the 313 participating patients, depending on the item, 10–12 had a missing response, and 4–39 gave an ‘unsure’ response. Of the 313 participating family members, depending on the item, 12–14 had a missing response, and 0–41 gave an ‘unsure’ response. Dyads with missing or ‘unsure’ responses were excluded from calculations of agreement on values. Crude agreement varied widely between items, with a range of 19.9% (item 8: How important is it that I/s/he respect the wishes of other family members regarding my/his/her care?) to 71.7% (item 1: How important is it that I/s/he be comfortable and suffer as little as possible?) (table 1). The average crude agreement for these eight items was 41.6%. In a sensitivity analysis in which we measured approximate agreement, we calculated the proportion of patient–family dyads that differed in importance ratings by two or less (table 1): the proportion of dyads with ratings within two points or less of each other ranged from 41.1% (item 8) to 88.9% (item 1), with an average approximate agreement of 63.0%.
Patient–family agreement on preferences for life-sustaining treatment
The distribution of patient–family dyad responses for medical treatment preferences are displayed in table 4. We excluded dyads where at least one member had a missing response or responded ‘unsure’, leaving us with a total of 264 dyads for this analysis. Sixteen patients (5.2%) and six family members (2.1%) gave an ‘unsure’ response. In the majority of cases (n=241, 91.3%), dyads agreed on CPR preferences (ie, selecting CPR vs any other interventions without CPR); kappa was 0.60 (95% CI 0.45 to 0.75) indicating good agreement. Within the 31 dyads in which the patient preferred to receive CPR, crude agreement was 67.7%. Within the 233 dyads in which the patient preferred not to receive CPR, crude agreement was higher (94.4%). However, when considering all five options for life-sustaining treatment on the spectrum from comfort-oriented care to use of all possible measures to prolong life, crude agreement between patients and family members fell to 56% (kappa 0.39 (95% CI 0.31 to 0.47), indicating a fair level of agreement). The most common sources of disagreement occurred because of differences in treatment preferences at a more detailed level than the simple distinction between CPR or no CPR.
Patient versus family member preferences for life-sustaining treatment
Predictors of patient–family agreement on preferences for life-sustaining treatment
In our multivariate analysis, four variables were independent predictors of patient–family member agreement on preferences for life-sustaining treatment (table 2): patient-reported education of elementary school or less was associated with significantly lower odds of agreement in comparison to those reporting postsecondary education (OR 0.21, 95% CI 0.07 to 0.66, p=0.005); patient-endorsed Catholic faith was associated with significantly lower odds of agreement versus those who endorsed no religious affiliation (OR 0.17, 95% CI 0.04 to 0.10, p=0.016); and patient self-reported ratings of being ‘fit’ on the Frailty Scale was associated with lower odds of agreement in comparison to those who rated themselves as being ‘frail’ (OR 0.50, 95% CI 0.26 to 0.96, p=0.038). Lastly, dyads in which the family caregiver was a spouse or partner had higher odds of agreement than those with other types of relationships (eg, daughter/son) (OR 3.84, 95% CI 1.51 to 9.81, p=0.005). Although the odds of agreement was higher among patients with an advance directive or living will, this association did not remain statistically significant in the multivariate analysis (OR 1.53, 95% CI 0.83 to 2.82, p=0.18), and there was no suggestion of association between agreement and previous engagement in any of the other advance care planning activities such as designation of a surrogate decision maker or previous experience in EOL decision making.
Discussion
In this multisite study of advance care planning practices conducted at 16 Canadian hospitals, we used a validated questionnaire to understand patient–family agreement on values and preferences for life-sustaining treatment. Although family members are required to represent the patient’s values and preferences when acting as surrogate decision makers, family members’ own values and preferences for life-sustaining treatment may undermine this requirement when at odds with patients’ wishes. As such, this study focused on the extent of agreement of family members’ own values and preferences for life-sustaining treatment, when compared with those of patients. We found that, despite over three quarters of patients and family members reporting previous discussions about preferences for care, there was generally low to moderate agreement in values about life-sustaining treatment and good agreement on CPR preferences but only fair agreement when looking at more detailed options for use or non-use of life-sustaining treatments. This is noteworthy, as previous work by our group has shown that, of the 76.3% of patients who had thought about life-sustaining treatment, a majority (92.0%) had discussed their preferences with family member(s).9 Our current study suggests that these conversations are not necessarily leading to agreement in values and preferences for life-sustaining treatment between patients and family members.
In reality, patients and family members are often presented with complex decisions, so understanding a patient’s goals of care from a values-based perspective is critical to guide decision making about the more nuanced but commonplace decisions about scope of treatment, beyond the question of ‘CPR vs no CPR.’ There are often unanticipated clinical scenarios and multiple factors that preclude the ability to plan ahead for each treatment decision that may be encountered in the future.16 As such, it is preferable to consider patients’ values, including what outcomes are hoped for or which are to be avoided if possible, and allow this to guide and prepare family members’ for future surrogate decision-making activities.
Our finding of only moderate agreement in values and preferences for life-sustaining treatment does not necessarily tell us about family members’ future ability in representing patients’ wishes. For example, although family members may have discordant preferences, they may be able to separate these preferences from their substituted decision-making role. Unfortunately, previous studies suggest the contrary. Fagerlin et al 5 found that surrogates’ own preferences were a significant predictor of accuracy of representation of patients’ wishes.4 Using the SUPPORT study data, Marks and Arkes4 estimated that, in 88.4% of cases where surrogates incorrectly predicted patient preferences for care, categorised as ‘extend life’ versus ‘relieve pain’, it was because of surrogate ‘projection’ of their own wishes onto the patient—their difficulty separating their own values from those of the patient. There is also the potential that family members who have discordant views from their seriously ill loved one may experience significant distress when making decisions that honour patients’ previously stated values. This is not an insignificant issue considering how emotionally invested family members are likely to be when it comes to decision making about life-sustaining treatments for a loved one.
In our exploratory multivariate analysis, we found that spousal relationships, higher patient education level and frailty are predictive of patient–family agreement on preferences for life-sustaining treatment, whereas endorsing Catholic faith appears to be associated with lower odds of agreement. Long-term spousal relationships may intuitively result in similarities in values and preferences compared with parent–child or other relationships, so this particular finding may not be surprising. Similarly, increased frailty may motivate patients and families to engage in advance care planning and thus lead to higher levels of agreement, and higher education levels may be associated with higher health literacy and recognition of the importance of engaging in advance care planning with loved ones. In our multivariate analyses, we found that previous engagement in advance care planning activities was not independently associated with higher odds of agreement on preferences for life-sustaining treatment between patients and family members. This suggests that current advance care planning activities are not achieving alignment between patients and family members in values and preferences for life-sustaining treatment and that there is a need to understand how to improve the effectiveness of such activities.
Strengths and limitations
The strengths of this study include the large sample size of hospitalised, seriously ill older adults from multiple sites of recruitment across Canada. In addition, a ‘real-time’ assessment was conducted, where participants were asked about their preferences as it related to current medical issues as opposed to hypothetical scenarios. Our study has several limitations. First, a participation rate of 56%, enrolment of a predominantly Caucasian, Christian population and exclusion from our analyses of cases with missing or ‘unsure’ responses are all factors which may limit the generalisability of our findings. Second, we asked family members for their own values and preferences for life-sustaining treatment, not what they thought the patient would want. Family members’ awareness of any differences between these two views is an important question that warrants further research. Finally, our questionnaire elicited values about the use of life-sustaining treatment using eight survey items that were only partially informed by a previously developed EOL Values Scale by Winter et al.13 14 The psychometric properties of the EOL Values Scale cannot be assumed to be transferable to the items in our survey that asked about values. However, we developed our survey items using input from content experts and pilot testing of our survey with patients and family members. Therefore, we conclude that the questions about values in our survey do have face and content validity.
Future directions
Given our findings of discrepancies in values and preferences for life-sustaining treatment within patient–family dyads, clinicians need to be mindful of and explore these differences during advance care planning. More can be done to prepare family members or remind them of their role in representing the patient’s wishes, not their own, when patient becomes incapacitated. Specifically, communication interventions that focus on values-based advance care planning discussions may be of benefit. The Serious Illness Care Programme is one such intervention currently being studied within a cluster-randomised controlled trial for patients with advanced cancer, and this multicomponent intervention includes a communication tool to facilitate values-based discussions between patients and family members.17 Other studies have found the use of video decision aids helpful in eliciting preferences and improving confidence in decision making about life-sustaining treatment for patients with cancer and dementia and for nursing home residents.18–20 Future studies could use similar methods but include family members with a focus on tools to support values-based advance care planning conversations between patient–family dyads.
Conclusions
Family members are frequently called on to make medical treatment decisions for seriously ill family members who cannot speak on their own behalf. Our study has found only moderate agreement between patients and family members with respect to values and preferences for life-sustaining treatments, beyond CPR preferences alone. This has implications for how clinicians help patients and their family members prepare for surrogate decision making and the dissonance family members may feel in attempting to honour their loved ones wishes for care at the end of life.
Acknowledgments
The authors thank the research coordinators at the following sites for their support conducting the study: Carol Mantle (Hamilton General Hospital); Christine Welsh (Ottawa Civic and General Hospitals); Lauren Ogilvie, Gail Gonyea, Lisa Sinclair, Tess Topor (Calgary Zone Hospitals, Alberta Health Services); LeeAnn Esau, Sarah Borland, (Chinook Regional Hospital) Pat Porterfield, Sandy MacDonaldis (Vancouver General Hospital); Reihbar Hans (Burnaby and Royal Columbian Hospitals, British Columbia); Betty Jean Ashley, Janice Palmer, Lucia Milosavljevic (St. Paul’s Hospital, Vancouver).
References
Footnotes
Contributors AA, JY, DK and JS were involved in the design of this study. DK, JY and JS were involved in data collection. AA, SG, AD, JY and DK were involved in data analysis. AA, JY, JS, SG and AD were involved in data interpretation, and all authors were involved in manuscript writing.
Ethics approval All study sites including Queen’s University, McMaster University, University of Calgary, Fraser Health, Universite de Sherbrooke.
Provenance and peer review Not commissioned; externally peer reviewed.