Assessing Family Members' Satisfaction with Information Sharing and Communication during Hospital Care at the End of Life

Abstract

Context:

Despite the fact that most deaths occur in hospital, problems remain with how patients and families experience care at the end of life when a death occurs in a hospital.

Objectives:

(1) assess family member satisfaction with information sharing and communication, and (2) examine how satisfaction with information sharing and communication is associated with patient factors.

Methods:

Using a cross-sectional survey, data were collected from family members of adult patients who died in an acute care organization. Correlation and factor analysis were conducted, and internal consistency assessed using Cronbach's alpha. Linear regression was performed to determine the relationship among patient variables and satisfaction on the Information Sharing and Communication (ISC) scale.

Results:

There were 529 questionnaires available for analysis. Following correlation analysis and the dropping of redundant and conceptually irrelevant items, seven items remained for factor analysis. One factor was identified, described as information sharing and communication, that explained 76.3% of the variance. The questionnaire demonstrated good content and reliability (Cronbach's alpha 0.96). Overall, family members were satisfied with information sharing and communication (mean total satisfaction score 3.9, SD 1.1). The ISC total score was significantly associated with patient gender, the number of days in hospital before death, and the hospital program where the patient died.

Conclusions:

The ISC scale demonstrated good content validity and reliability. The ISC scale offers acute care organizations a means to assess the quality of information sharing and communication that transpires in care at the end of life.

Introduction

Despite the fact that the home is the preferred site of care and death for most people, most deaths continue to occur in hospital settings.^1,2 Awareness of the large number of deaths occurring in acute care settings has generated investigations into the quality of care hospital patients receive at the end of life. In the United States, the landmark Study to Understand Prognoses and Preferences for Outcome and Risks of Treatment (SUPPORT) heightened interest in the need to improve hospital care for dying patients.³ The SUPPORT findings revealed poor communication between physicians and patients on preferences for care, poor pain and symptom management, and the continuation of aggressive therapy without sufficient attention to palliation in the face of almost certain death.³ The shortcomings identified in hospital care by SUPPORT continue to be documented in more recent studies.^4–7

An important aspect of evaluating quality of end-of-life care is whether patients and families are satisfied with care; however, the lack of a widely accepted conceptual framework for satisfaction at the end of life complicates efforts to assess satisfaction with care. When satisfaction is assessed, instruments may either assess satisfaction globally or may focus on specific elements of care.⁸ One specific attribute that has been identified by patients and families as representing quality care at the end of life is that patient and family treatment preferences and decisions will be assessed and respected.⁹ This attribute is reflective of the quality of the process of care at the end of life, namely, information sharing and communication among providers, patients, and family members. The primary objective of this paper is to describe the development of an adapted measure that assesses this element of hospital care at the end of life. A second objective is to report what patient factors are associated with family members' satisfaction with information sharing and communication in care at the end of life. It is anticipated that this work will facilitate an evidence-based approach for an organization to assess its performance on this attribute of care.

Methods

Setting

Hamilton Health Sciences, located in southern Ontario, Canada, includes three tertiary acute care academic teaching hospital sites: Hamilton General Hospital, Juravinski Hospital/Juravinski Cancer Centre, and McMaster University Medical Centre/McMaster Children's Hospital. Approximately 1600 people die annually within this organization. Responding to the evidence that patients and families facing terminal illness want to engage with health care providers in proactive planning of end of life,¹⁰ the organization launched an initiative to facilitate a proactive and collaborative approach to end-of-life care planning. The initiative involved several components that included a document based on emerging best practice in the United States entitled the Physician's Orders for Life-Sustaining Treatment (POST),¹¹ as well as developing staff capacity through education to support patients and their families through the end-of-life care planning process. The initiative and POST also emphasized a change in language away from do-not-resuscitate to acknowledging patient wishes for comfort measures while allowing natural death.

Sample

As part of the evaluation of this initiative, a cross-sectional survey of family members was conducted in 2010. The purpose of the survey was to record family members' satisfaction with aspects of care at the end of life. The questionnaires were sent by mail to the listed next-of-kin for each patient who died while hospitalized between January 1, 2010 and December 31, 2010. Two to three weeks after the death, the next of kin were sent a pre-notice letter describing the study and notifying that a survey would be sent in two weeks. Two weeks after the pre-notice letter, the survey was sent with an information/consent letter about the study and a self-addressed, stamped envelope to return the completed survey. If no response was received within four weeks, a reminder letter and second questionnaire were mailed. Four weeks after mailing of the second survey, a final reminder and survey were sent.

Measure

Two previously published surveys evaluating end-of-life care were identified as potential instruments to be used for this study: CANHELP (Canadian Health Care Evaluation Project-Caregiver Questionnaire) family version¹² and FAMCARE.¹³ CANHELP is a 38-item instrument developed for use in community and acute care settings, with both patient and family member versions. Respondents are asked to assess both perceived importance and satisfaction with care during the four weeks prior to completing the questionnaire. Domains assessed in the family version include relationships with doctors, characteristics of doctors and nurses, illness management, communication and decision making, family member involvement, and well-being. The instrument also includes several open-ended questions. FAMCARE is a 20-item instrument developed for completion by caregivers with a family member who has cancer. While the instrument is said to assess four domains: information giving, availability of care, physical care, and pain control, research suggests that the instrument is unidimensional and assesses “general care.”¹³ The research team determined that neither instrument was suitable to meet the information needs of the present study, which had a focus on provider engagement with patients/families' engagement with end-of-life care planning. The investigators also sought an instrument that was easy to administer and complete, using a mailed survey approach.

On the basis of this decision, the study investigators developed a scale to address the information needs of the present study. Both CANHELP and FAMCARE were reviewed for the identification and selection of suitable items for inclusion in the scale. In this study, only perceived satisfaction with information sharing was elicited; rating the importance of the given issue, as was done in the CANHELP survey, was excluded. To suit a bereavement survey approach, wording of the FAMCARE items was changed from the present tense to the past tense (e.g., “speed with which symptoms were treated” from “are treated”). In addition, the words “doctor” and “nurse” were changed to “health care team” to reflect the interprofessional approach to care in hospital settings. A pediatric version was also developed, in which the use of the term “the patient” was changed to “your child.” Response scales for all items ranged from 1 (not at all satisfied) through to 5 (completely satisfied). The final set of items was reviewed and approved by domain experts. Domain experts included an acute care palliative care physician, an acute care pediatrician, and a hospital ethicist.

Statistical analysis

To ensure a homogeneous study sample, data analysis focused on the adult version of the instrument. Data analysis for this study consisted of two phases. The first involved examining the measurement properties of the new scale. The second phase analyzed the association between family member satisfaction with information sharing and communication, and patient variables.

Following data collection, a correlation analysis was conducted to identify highly correlated (r ≥0.80) items. Highly correlated items were considered to be conceptually redundant, and were reviewed by domain experts to determine which of the two highly correlated items would be removed along with items considered conceptually irrelevant to the scale. Exploratory factor analysis with principal axis factoring was then used to identify the factor structure among the items. Principal axis factoring was chosen for the factor solution because it does not include measurement errors¹⁴ and made the most conceptual sense as described by Pett and colleagues.¹⁵ Adequacy of extraction and factor retention criteria included the size of the eigenvalues (>1.0) and residuals, scree plot, and factor interpretability and usefulness.^14,15 In the extraction phase, a factor loading of at least 0.40 was used to include an item in a factor.¹⁵

Internal consistency of the factor was estimated using Cronbach's alpha, Cronbach's alpha-if-item-deleted, and corrected-item-total correlations.¹⁶

Means and standard deviations were used to describe patient characteristics. Bivariate analysis including t-tests, chi-square, and analysis of variance were used to determine which of the patient variables were associated with respondent satisfaction with information sharing and communication. The independent contribution of predictor variables was then assessed using linear regression. For the linear regression, all statistically significant variables (according to the bivariate analyses) were used. Stata statistical software version 11.2 (StataCorp., College Station, TX) was used for statistical computations with a significance level of 0.05.

Results

Patient characteristics

During the survey period, 1216 questionnaires were mailed out to the adult study sample; 529 questionnaires were completed and returned to the study team (43% response rate). One-half of survey respondents were spouses of the deceased (51%). The average age of patients at the time of death was 75.0 years (SD 13.8). The average length of stay in the hospital for the decedent population was 14.5 days (SD 26.9). Other characteristics of the population are shown in Table 1.

Table 1.

Description of Patient Population

Characteristic		Test and P value for association with overall summed ISC score
Mean age (SD)	75.7 (13.8)	r=0.0444, 0.308^a
Female % (n)	47.3 (250)	t=2.184, 0.029
Mean number of days in hospital before death (SD)	14.5 (26.9)	r=−0.162, <0.001
Relationship of next of kin, % (n)
Spouse	53.5 (283)	F=0.40, 0.473^b
Child (including in-law children)	33.3 (176)
Parent	2.5 (13)
Other relative	10.8 (57)
Hospital in which patient died, % (n)
Hospital 1	19.5 (103)	F=0.75, 0.473
Hospital 2	42.2 (223)
Hospital 3	38.4 (203)
Hospital program where patient died, % (n)
Surgery^c	10.6 (56)	F=1.54, 0.188
Medicine^d	35.5 (188)
Critical Care (ICU/CCU)	25.7 (136)
Oncology	18.0 (95)
Cardiology	10.2 (54)

Pearson correlation test.

F-value is reported from an ANOVA test.

Includes general, cardiovascular, and neurosurgery, and deaths in orthopedic unit.

Includes one death in burn unit.

Development of the information sharing and communication (ISC) scale

The initial version of the survey sent to family members contained 23 items. Following correlation analysis and dropping of redundant or conceptually irrelevant items, 7 items remained for factor analysis. We used principal component methods for factor extraction. In the factor analysis, only one factor emerged with eigenvalue equal to 5.34, accounting for 76.3% of the variance. The factor was identified as information sharing and communication. The internal consistency of the questionnaire was high (Cronbach's α=0.95). Corrected-item-total correlation for the factor ranged from r=0.73 to r=0.87, which was in the acceptable range. Finally, for each item the Cronbach's alpha-if-item-deleted from the scale was also analyzed no further deletion of any items was required. Items that comprise the final instrument are presented in Table 2.

Table 2.

Description and Respondent Scores for Each Item in the Survey Instrument

Items	n	Mean score^a (SD)
1. Information provided about the patient's prognosis.^b	523	3.9 (1.2)
2. Information given about side effects of treatments.	490	3.8 (1.2)
3. Information given about how to manage the patient's symptoms.^b	481	3.8 (1.2)
4. Information given about the patient's test results.	505	3.9 (1.2)
5. Consistency of information about the patient's condition from all the doctors and nurses looking after him/her.	520	3.8 (1.3)
6. Information provided about what to expect at the end stage of the patient's illness (e.g., in terms of symptoms and comfort measures).	503	3.9 (1.3)
7. The control of the patient's physical symptoms (e.g., pain, shortness of breath, nausea).	517	4.0 (1.1)
8. When the team initiated end-of-life care discussions. (“Were you given enough time to plan for the patient's death?”)^b	492	3.8 (1.4)
9. The speed with which symptoms were treated.	518	4.0 (1.2)
10. The way tests and treatments were performed.	509	4.0 (1.1)
11. The way tests and treatments were followed up by the health care team.	509	4.0 (1.2)
12. The patient's pain relief.	507	4.1 (1.1)
13. Availability of the health care team to the patient.	510	4.1 (1.1)
14. Availability of the health care team to you the family member.	514	4.0 (1.2)
15. The health care team's knowledge of the patient's wishes for end of life care.^b	476	4.1 (1.2)
16. How the health care workers worked together as a team to look after the patient.	515	4.2 (1.1)
17. How closely the health care team followed the wishes of the patient.	473	4.2 (1.1)
18. Coordination of the care of the patient.	507	4.1 (1.2)
19. How your questions were answered by the health care team.^b	516	4.1 (1.1)
20. Family conferences held to discuss the patient's illness.	484	3.9 (1.3)
21. The way the family was included in treatment and care decisions.	503	4.0 (1.2)
22. The quality of the discussions with the health care team about the use of life-sustaining technologies (e.g., CPR or cardiopulmonary resuscitation or feeding and breathing tubes).^b	480	4.1 (1.3)
23. Your ability to talk comfortably with the patient about his or her illness, dying, and death.^b	415	3.7 (1.4)
Overall score^c		3.9 (1.1)

Each question scored on a 5-point Likert scale (1=not at all satisfied, 5=completely satisfied).

Items comprising the new ISC scale.

Calculated using only the items comprising the new ISC scale.

Respondent perceptions on end-of-life care

Family member satisfaction with information sharing and communication was computed by averaging the total score of the items from the factor that emerged. Overall, family members were satisfied with how information was shared and how communication transpired during the care their family member received at the end of life; the mean total satisfaction score was 3.9 (SD 1.1) (Table 2). Table 3 summarizes the linear regression model for the dependent variable, family member satisfaction with information sharing and communication. The model demonstrates that a longer length of stay in the hospital was associated with lower reported satisfaction with information sharing and communication during the care provided at the end of life. A gender difference was also noted where family members of a patient who was male were less satisfied with information and communication (female, x=4.0, SD=0.90; male, x=3.8, SD=1.2) Respondents who had a family member die within the critical care program reported higher satisfaction on information and communication during care at the end of life.

Table 3.

Predictors of the Overall Summed ISC Score

Variable	Regression coefficient (95% confidence interval)	P-value
Gender	0.23 (0.04, 0.41)	0.015
Program
Surgery^a	–^b
Medicine^c	0.13 (−0.17, 0.43)	0.40
Critical care (ICU/CCU)	0.39 (0.07, 0.70)	0.02
Oncology	0.21 (−0.12, 0.55)	0.21
Cardiology	0.18 (−0.20, 0.56)	0.35
Days in hospital before death	−0.006 (−0.009, −0.003)	<0.001
Constant	3.87 (3.60, 4.15)	<0.001

Includes general, cardiovascular and neurosurgery, and deaths in orthopedic unit.

Reference category.

Includes one death in burn unit.

Discussion

It is considered that family satisfaction is an important criterion to assess the quality of care received at the end of life. For researchers and policymakers, assessing satisfaction with care may be used as an evaluative outcome. From an administrative perspective, assessing satisfaction has become important as a tool in the era of continuous quality improvement, where quality of care is viewed as reducing patient “burden.”

The present study describes the development of a seven-item instrument that generates a single total score summarizing family members' satisfaction with information sharing and communication that transpires during hospital care at the end of life. Exploratory factor analysis identified a single factor, described as information sharing and communication, that explained 76.3% of the variance. The questionnaire also demonstrated excellent internal consistency (Cronbach's α=0.95). Focus of the instrument on a specific attribute of quality care at the end of life legitimates the use of a single total score.

The lack of a widely accepted conceptual framework for satisfaction complicates efforts to measure satisfaction with care. Recent efforts have focused on identifying the patient and family perspective on the quality of palliative care.⁹ In a recent metasynthesis conducted by Howell and Brazil,⁹ attributes of quality palliative care from the patient and family perspective were reviewed. The results identified several common attributes including the domain reflected in this study, “that treatment preferences and decisions will be assessed and respected.”

The accuracy of measuring family satisfaction with care at the end of life can be affected by a number of factors. A key factor is the quality of the instrument used, in terms of validity and reliability. In this study, reliability of the ISC scale was examined by Cronbach's alpha, which assesses internal consistency; while validity was examined through content validity (i.e., face validity), two closely related concepts. Reproducibility (test-retest) represents a further future step in the development of this scale. As well, further work should be directed toward criterion or construct validity. However, the absence of a well developed conceptual model of satisfaction does challenge validation assessment of these two elements.

There are numerous factors that can influence a family member's response on a survey instrument, including social desirability and acquiescence bias. Recognizing the potential limitations of using a quantitative approach to assess satisfaction, qualitative approaches have been recommended.⁸ In response to these recommendations, it is suggested that the current instrument be augmented with a section to solicit open-ended comments from respondents on perceived quality of information sharing and communication.

The results of the study revealed that most family members were generally satisfied with the information sharing and communication making that occurred during hospital care at the end of life. High levels of satisfaction with care have been reported elsewhere.⁸ Low variation commonly noted in satisfaction, such as skewed distribution of responses and ceiling effects, was evident in this study. This observation in our findings reiterates the importance integrating open-ended questions that allow respondents to speak to their own experiences in their own words. Although the original instrument used in the study did not solicit open-ended comments, it was observed that a large number of respondents did under their own volition append narrative commentary to the questionnaire. Finally, caution must be applied to the interpretation of the results in recognition of a response rate of 43% which raises the question of the representativeness of the results.

This study revealed that family satisfaction with information sharing and communication was associated with patient variables gender and length of stay. Patient place of death in the hospitals also influenced perceptions on the quality of information sharing and communication at the end of life. At this point, efforts to explain these relationships would be speculative. While more in-depth investigation is required on these relationships, the findings reveal how perceptions of satisfaction are influenced by the unique context factors associated with processes of care.

There have been increasing efforts to improve the quality of care for the dying in acute care settings. Despite these efforts, persistent problems reported in hospital settings with do-not-resuscitate orders suggest that issues with information sharing and communication are still prevalent in acute care settings.⁶ The ISC scale demonstrated good content validity as well as reliability. The pediatric version of the ISC scale is likely to be equally as reliable and valid given how similar the instruments are; however, further evaluation would need to confirm this.

Footnotes

Acknowledgments

Funding was from the Hamilton Academic Health Sciences Organization (HAHSO) Innovation Fund.

Author Disclosure Statement

No competing financial interests exist.

References

Wennberg

. The Dartmouth Atlas of Health Care in the United States: Airport on the Medicare Program. Dartmouth, NH: Dartmouth Medical School, Center for the Evaluative Clinical Sciences, 1999; 180–181.

Wilson

, Truman

, Thomas

et al. The rapidly changing location of death in Canada 1994–2004. Soc Sci Med, 2008; 68:1752–1758.

The SUPPORT principal investigators:. A controlled trial to improve care for seriously ill hospitalized patients: The study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT) JAMA, 1995; 274:1591–1598.

Heyland

, Groll

, Rocker

et al.

End-of-life care in acute care hospitals in Canada: A quality finish?

J Palliat Care, 2005; 21:142–150.

Wright

, Ray

, Zhang

et al. Medical care and emotional distress associated with advanced cancer patients' end-of-life discussions with their physicians. J Clin Oncol, 2008; 26,15S:6505.

Yuen

, Reid

, Fetters

. Hospital do-not-resuscitate orders: Why they have failed and how to fix them. J Gen Intern Med, 2011; 26:791–797.

Zhang

, Wright

, Nilsson

et al. Associations between advanced cancer patients' end-of-life conversations and cost experiences in the final week of life. J Clin Oncol, 2008; 26,15S:9530.

Brazil

. Assessing care needs and satisfaction with health care delivery. Hudson

, Payne

. Family Carers and Palliative Care: Guide for Health and Social Care Professionals. Oxford: Oxford University Press, 2008; 113–130.

Howell

, Brazil

. Reaching common ground: A patient-family based conceptual framework on quality end-of-life care. J Palliat Care, 2005; 1:19–26.

10.

Heyland

, Cook

, Rocker

et al. CARENET. The development and validation of a novel questionnaire to measure patient and family satisfaction with end-of-life care: The Canadian Health Care Evaluation Project (CANHELP) Questionnaire. Palliat Med, 2010; 24:682–695.

11.

Heyland

, Dodek

, Rocker

et al. CARENET. What matters most in end-of-life care: Perceptions of seriously ill patients and their family members. CMAJ, 2006; 174:627–633.

12.

Hickman

, Sabatino

et al. The POLST (Physician Orders for Life-Sustaining Treatment) paradigm to improve end-of-life care: Potential state legal barriers to implementation. J Law Med Ethics, 2008; 36:119–144.

13.

Ringdal

, Jordhoy

, Kaasa

. Measuring quality of palliative care: Psychometric properties of the FAMCARE Scale. Qual Life Res, 2003; 12:167–176.

14.

Tabachnick

, Fidell

. Using Multivariate Statistics. Boston: Allyn and Bacon, 2001.

15.

Pett

, Lackey

, Sullivan

. Making Sense of Factor Analysis: The Use of Factor Analysis for Instrument Development in Health Care Research. Thousand Oaks, CA: Sage, 2003.

16.

Netemeyer

, Beardon

, Sharma

. Scaling Procedures: Issues, Applications. Thousand Oaks, CA: Sage, 2003.