Abstract
Background:
Understanding the factors that affect the congruence between preferred and actual place of death may help providers offer clients customized end-of-life care settings. Little is known about this congruence for cancer patients in receipt of home-based palliative care.
Objectives:
This study aims to determine the congruence between preferred and actual place of death among cancer patients in home-based palliative care programs.
Design:
A longitudinal prospective cohort study was conducted. Congruence between preferred and actual place of death was measured. Both univariate and multivariate analyses were used to assess the determinants of achieving a preferred place of death. From July 2010 to August 2012, a total of 290 caregivers were interviewed biweekly over the course of their palliative care trajectory from entry to the program and death.
Results:
The overall congruence between preferred and actual place of death was 71.72%. Home was the most preferred place of death. The intensity of home-based nursing visits and hours of care from personal support workers (PSWs) increased the likelihood of achieving death in a preferred setting.
Conclusions:
The provision of care by home-based nurse visits and PSWs contributed to achieving a greater congruence between preferred and actual place of death. This finding highlights the importance of formal care providers in signaling and executing the preferences of clients in receipt of home-based palliative care.
Introduction
Health system restructuring has resulted in a greater emphasis on home-based palliative care in contrast to institutional care. Home-based palliative care enables patients to be cared for at home, which is a familiar environment for patients and their families, and may also improve their quality of life.1,2 For terminally ill patients in receipt of home-based palliative care, knowing where they would like to die may also improve their comfort during their final stage of life. Home is often assumed to be the preferred place of death for patients3,4; however, not all patients prefer to die at home. 5 Knowing the patient's preferred place of death and the factors that hinder and facilitate its actuality would assist care providers in ensuring that such preferences are achieved when possible.
Previous studies have assessed the factors associated with achieving a preferred place of death.6–8 A variety of factors, such as higher social class, 9 hospice enrollment 10 and family support, 11 contribute to achieving a preferred place of death, whereas other variables, such as unclear prognosis 12 and lower patient education level, 13 decreased the likelihood of dying in a preferred setting. A systematic review summarized the three main factors that account for congruence between the preferred and actual place of death: environmental factors, illness-related factors, and individual factors. 14 Although some studies have explored the congruence between preferred and actual place of death through the use of cross-sectional data, there is a paucity of studies that have used longitudinal data. Dynamic variables, such as the intensity of use for palliative care services over the course of the palliative care trajectory, have not been examined to date and are most appropriately assessed longitudinally. Furthermore, little is known about the preference for place of death among cancer patients who receive home-based palliative care.
The aim of this study was to (1) understand preferences for place of death among cancer patients who were in receipt of home-based palliative care and (2) identify the factors associated with achieving congruence between those preferences and their actual place of death.
Methods
Design
This longitudinal prospective cohort study examined the preferences for place of death among palliative care patients receiving home-based health care services. Primary caregivers of patients were study participants, who were recruited from two home-based palliative care programs in Ontario, Canada. Potential participants were primary caregivers (family or friends) of cancer patients admitted to the palliative care program within the past 7 days, ≥18 years of age, and fluent in English. The programs provide palliative care services, including home-based physician visit services, nursing visit services, and supportive care from personal support workers (PSWs). The receptionist identified and telephoned potential eligible participants, and asked if they were interested in participating in the study. Written consents were mailed to and completed by participants who agreed to participate.
During the period from July 2010 to August 2012, caregiver participants were interviewed by telephone every two weeks from admission into the palliative care program until the death of the patient. To minimize recall bias and not to overburden participants, a two-week period for data collection was selected. Caregivers of patients were asked to recall the health care services used by the patient, the patient's functional status, and their own caregiver burden for the previous two weeks. It was postulated that it would be difficult for patients to participate in this study because of their deteriorating health over the course of palliative care trajectory, and accordingly, caregivers were interviewed. Both the University of Toronto and Mount Sinai Hospital ethics review boards approved this study (Reference no. 28971, Original Approval Date: July 9, 2013).
Measures
Two systematic review articles14,15 helped to guide the research framework. This framework allowed us to consider three sets of factors that may influence the congruence between the preferred and actual place of death: environmental factors, individual factors, and illness-related factors.
Information on each patient's gender, age, marital status, education, and deprivation score was obtained from regional databases for palliative care programs and home-care agency databases using the patient's unique health card number. Caregiver participant's gender, age, education, employment status, and relationship to the patient were collected during the first interview. In addition, information on a patient's actual place of death, whether living alone, and their comorbidity score were also obtained from regional databases for palliative care programs and home-care agency databases. The patient's attending palliative care physician collected the data on a patient's preferred place of death in the patient's home during their first home meeting with the patient. So, only the baseline preferred for place of death was reported by patients and this preference was used in the study. Although the patient's preferred place of death was recorded only at the baseline of the study, there is literature to suggest that preference for place of death may remain stable over the palliative care trajectory for the overwhelming majority of patients. 16 In addition, Advance Care Planning (ACP) or similar actions were not available for patients and their caregivers to consider at the time of the study. Consequently, the patient's preference for place of death was not influenced by the occurrence of ACP.
The Ambulatory and Home Care Record (AHCR) 17 (© Coyte and Guerriere 1998), a tool designed to measure health services utilization, was administered during each interview to collect data on health service use. The AHCR has been used in a variety of clinical settings and has been shown to yield valid and reliable data. 18 The agreement (kappa: 0.41–1.00) between participants' responses and administrative data was also reported. 18 Three common services provided by multidisciplinary home-based palliative care teams included home-based physician and nursing visits, and home-based PSW hours of care. 19 Data on these services were taken from the AHCR. 20 Both the propensity (whether the patient received the specific service or not) to use services and the intensity of using the services (frequency of service use) were reported by caregivers. The use of emergency visit services was also recorded during each interview.
The Caregiver Burden Scale in the End-of-Life Care (CBS-EOLC) 21 was used to assess caregiver burden. It is a 16-item questionnaire using a 4-point Likert scale. The total burden ranges from 16 (low burden) to 64 (high burden). Appropriate reliability of the scale (Cronbach's alpha = 0.95) and good levels of convergent validity with fatigue (r = 0.69) and depression (r = 0.54) 21 have been demonstrated.
The Palliative Performance Scale (PPS) 22 was employed to measure the functional status of patients. The reliability test of PPS reported a high intra-class correlation coefficient (0.93–0.96).23,24 The PPS score ranges from 0%, which represents death, to 100%, which represents full ambulation and the absence of disease.
Data analysis
Descriptive statistics were used to represent patients' and caregivers' demographic characteristics. The characteristics of patients who achieved their preferred place of death and those who did not were presented using cross-tabulations. The primary outcome was achieving a preferred place of death (i.e., the congruence between preferred and actual place of death). The congruence was presented in terms of frequency and percentage. The extent of congruence was assessed using a kappa statistic. Logistic regression was employed to analyze the determinants of achieving a preferred place of death. Both univariate and multivariate analyses, which were based on interview-level (longitudinal) data, were used to analyze the factors associated with congruence. A backward stepwise regression approach was adopted in the multivariate analysis. Through a backward stepwise regression analysis, the final multivariate model included all variables that were statistically significant at the alpha = 0.1 level. 25 All analyses were performed using Stata (Stata version 13.0 for Mac, StataCorp LP, College Station, TX).
Results
There were 805 eligible participants identified for the period from July 2010 to August 2012. Eighty-eight potential participants were under extreme stress and were excluded from the study. Five hundred and fifty-two (77.0%) caregivers of the remaining 717 eligible participants were contacted, 367 (66.5%) of those agreed to receive further study information, and 341 (92.9%) of those who received further information agreed to participate in the study. However, 14 participants were ineligible at the first interview because the patient had died (n = 9), was hospitalized (n = 4), or had moved (n = 1). 25 A further 37 participants were excluded because data on their preferred or actual place of death were missing. Consequently, our analyses were based on a group of 290 caregiver participants that yielded a total of 1582 interviews over the course of palliative care trajectory. Caregivers participated in at least one interview and the average number of interviews per caregiver was 5.5.
Patient and caregiver characteristics
Table 1 presents the descriptive statistics for patients and caregivers and the distribution of whether they achieved a preferred place of death by each of the variables. Most patients were aged 65–84 years, with a mean age of 72.4 years. More than half of the patients were female (53.79%) and married (59.66%). Most of the patients lived together with others (83.79%). The majority (67.24%) of caregivers were women. A significantly greater proportion of patients who achieved a preference of place of death lived with others.
Patient and Caregiver Characteristics (n = 290)
Over the course of palliative care trajectory.
PPS, Palliative Performance Scale; PSW, personal support worker; SD, standard deviation.
Congruence between the preferred and actual place of death
The actual and preferred place of death and their congruence were shown in Table 2. The overall congruence rate was 71.72%, with an overall kappa value of 0.527, indicating moderate agreement. 26 Among the 290 decedents, home was the most preferred place of death (65.52%), followed by hospital (16.89%), hospice (16.21%), and finally, a long-term care (LTC) facility (1.38%). Patients who preferred to die in a hospice were the most likely group to have their wishes met (85.11%).
Congruence between the Preferred and Actual Place of Death among All the Decedent (n = 290)
The overall congruence was 71.72% with an overall kappa statistic of 0.527.
Values in bold and italic illustrate number and proportion of decedents who achieved congruence.
LTC, long-term care.
Factors associated with achieving a preferred place of death
Table 3 reports the results of the univariate analysis. Patient's and caregiver's age, deprivation score, caregiver burden, comorbidity score, and PPS were regarded as continuous variables in the regressions. Thirteen variables were associated with achieving a preferred place of death. Those variables included patient's gender, marital status, education, living arrangement, PPS scores, caregiver's employment status and education, their relationship to the patient, caregiver burden, the intensity of use of home-based physician visits, nursing visits and PSW hours of care, and the propensity to use home-based physician visits.
Univariate Odds Ratios from a Logistic Regression for Likelihood of Achieving a Preferred Place of Death (n = 1582)
Propensity: For each interview, propensity was 1 if the service was received in the interview.
Intensity: The number of services that patient used in each interview.
Patient's and caregiver's age, deprivation score, caregiver burden, comorbidity score, and PPS were regarded as continuous variables in the regressions.
LCI, lower confidence intervals; OR, odds ratio; UCI, upper confidence intervals.
Table 4 shows the results of the multivariate analysis. The backward stepwise regression analysis revealed four variables that affected the congruence between the preferred and actual place of death. Increasing the intensity of home-based nursing visits (odds ratio [OR] 1.02; 95% CI: 1.00–1.04) and hours of PSW care (OR 1.09; 95% CI: 1.01–1.18) increased the likelihood of achieving a preferred place of death. Patients who had higher PPS scores (OR 1.02; 95% CI: 1.01–1.04) were more likely, and those who were divorced, separated, or widowed (OR 0.45; 95% CI: 0.36–0.56) were less likely, to achieve their preferred place of death.
Multivariate Odds Ratios from a Logistic Regression for Likelihood of Achieving a Preferred Place of Death (n = 1582)
Propensity: For each interview, propensity was 1 if the service was received in the interview.
Intensity: The number of services that patient used in each interview.
Patient's and caregiver's age, deprivation score, caregiver burden, comorbidity score, and PPS were regarded as continuous variables in the regressions.
We also conducted a patient-level analysis for our study and found that the significance of the propensity and intensity variables was robust to the level of analysis, whether interview based or patient based.
Discussion
This study assessed the congruence between preferred and actual place of death among cancer patients who were in receipt of home-based palliative care and identified the variables that were associated with congruence. Several dynamic variables, such as the intensity of home-based palliative care service use over the course of the palliative care trajectory, were examined in this study. Herein, we compare our results with previous studies. Our findings are compared with those studies that had the primary purpose of examining the congruence between preferred and actual place of death. The majority of the previous studies used cross-sectional data, whereas we used longitudinal data and an interview-based analysis.
Our study found that home was the most preferred place of death for patients, which was consistent with previous studies.7,27,28 In our study, the overall congruence between preferred and actual place of death was 72%. This rate is higher than that of 67% reported in a previous study. 29 However, the patients in their study did not receive care through a home-based palliative care program, and the proportion of patients achieving a preferred home death was lower than that in our study. A recent Dutch study by de Graaf et al. that focused on home-based hospice care (including general practitioners, district nurses, trained volunteers, and a hospice team) reported that 91% of patients achieved a preferred place of death, 30 and the congruence rate was higher than that in our study. They found that hospice assist at home, a collaboration between general practitioners, district nurses, trained volunteers, and a hospice team, helps patients achieve their preferred place of death.
In this study, patients who were divorced, separated, or widowed were less likely to achieve a preferred place of death than those who were married. This might be due to the spouse being there to care for the patient, advocate on behalf of the patient and to actualize the patient's preferred place of death. Our results were consistent with a study conducted in Denmark, which included end-stage cancer patients and also found that patients who were married were more likely to achieve a preferred place of death. 31 Conversely, our results were at variance with a German study, which did not find an association between marital status and the congruence. However, the German study was a cross-sectional study that was based on a random sample of 5000 inhabitants of Rhineland-Palatinate, included cancer and noncancer patients, and did not focus on a palliative program. 32 The German study found that living together with a relative contributed to the congruence between preferred and actual place of death, which stressed the importance of informal care in facilitating death in a patient's preferred setting.
The intensity of home-based nursing visits and hours of PSW care increased the chance of achieving a preferred place of death in this study. We did not identify any previous studies that examined the impact of the intensity of using home-based nursing services on achieving a preferred place of death. However, we did not find any evidence that the propensity of both home-based nursing visits and PSW hours of care affected congruence. Conversely, a study by Burge et al. 28 did report that the propensity to use nursing services at home contributed to the achievement of a preferred home death; however, they did not examine the effect of the intensity of using nursing care. We did not identify any studies that examined the impact of the propensity to use and the intensity of home-based PSW hours on achieving a preferred place of death.
Patients with higher PPS scores were more likely to achieve a preferred place of death. Perhaps patients with higher scores have better physical function and thus have more time to discuss their preferences with care providers and their family. We did not identify any other study that assessed the impact of PPS scores on achieving a preferred place of death. However, previous studies reported that inability to control symptoms decreased the congruence between preferred and actual place of death,33,34 which is consistent with what the PPS scores in our study represent.
Several study limitations are to be noted. First, generalizing the study findings to other palliative care settings may be difficult as this study was based on data from two home-based palliative care programs in Toronto, Canada. However, patients participating in this program were diverse in clinical, demographic, and ethnic background, which helps to improve its generalizability. Second, some of the data used were drawn from telephone interviews with caregivers, which may result in a potential for social desirability and recall bias. 35 However, because the psychometric properties of the study instruments have met acceptable standards, we believe these potential biases are small. Third, information concerning variations in preferences for place of death over the palliative trajectory was not recorded in this study as preferences were only asked during the first home meeting with patients after entry into the palliative care program. Consequently, we were unable to assess whether patients changed their preferences for place of death and how such variations may have affected the results. However, one previous study has reported that the overwhelming majority (80%) of patients have stable preferences for their place of death as their illness progressed. 16 So we believe our results are robust to this consideration.
Understanding the factors that affect congruence between preferred and actual place of death may help patients and their families achieve their goals and help providers offer clients customized end-of-life care settings. Our findings strongly support the importance of home-based palliative care services, especially home-based nursing services and PSW services, which may help patients die in their preferred placed of death. Our study also found that most patients preferred a home death. This finding suggests that if support for patients desire to achieve a preferred place of death is a policy priority, health mangers and policy makers may help support families by ensuring timely access to and sufficient provision of home-based nursing and PSW services.
Conclusions
This longitudinal prospective cohort study examined the congruence between preferred and actual place of death for cancer patients who were in receipt of home-based palliative care, and identified the factors associated with achieving a preferred place of death. The overall congruence rate was 72%. Home was the most preferred place of death. Our findings emphasize the importance of the provision of formal care services, such as home-based nursing visits and hours of PSW care, that support patients to achieve their preferred place of death. These findings have important implications for home care managers and policy decision makers in their design of successful patient-centered home-based palliative care programs.
Footnotes
Funding Information
This study was funded by National Natural Science Foundation of China (grant no. 71904011).
Author Disclosure Statement
No competing financial interests exist.