Location of Clinician-Family Communication at the End of Life in the Pediatric Intensive Care Unit and Clinician Perception of Communication Quality

Abstract

Background: Communication between clinicians and families of dying children in the pediatric intensive care unit (PICU) is critically important for optimal care of the child and the family.

Objective: We examined the current state of clinician perspective on communication with families of dying children in the PICU.

Design: Prospective case series over a 15-month study period.

Setting/Subjects: We surveyed nurses, psychosocial staff, and physicians who cared for dying children in PICUs at five U.S. academic hospitals.

Measurements: Clinicians reported on the location of communication, perceived barriers to end-of-life care, and rated the quality of communication (QOC).

Results: We collected 565 surveys from 287 clinicians who cared for 169 dying children. Clinicians reported that the majority of communication occurred at the bedside, and less commonly family conferences and rounds. Ten barriers to care were examined and were reported with frequencies of 2%–32%. QOC was rated higher when the majority of conversations occurred during family conferences (p = 0.01) and lower for patients of non-white race (p = 0.03). QOC decreased when 8 of the 10 barriers to care were reported.

Conclusions: When a child is dying, clinicians report that communication with the family occurs most frequently at the child's bedside. This has important implications for future ICU communication research as the majority of previous research and education has focused on family care conferences. In addition, findings that QOC is perceived as lower for non-white patients and when clinicians perceive that barriers hindering care are present can help direct future efforts to improve communication in the PICU.

Introduction

Communication between clinicians and families is central to high-quality care of dying patients in the pediatric intensive care unit (PICU).^1–6 High-quality communication can aid complex and difficult decision making at the end of life, can help to improve the quality of medical care provided, and can reduce the risk of post-traumatic stress disorder (PTSD) and other mental illnesses in family members after the death of their child.^7–12

Communication is a broad term in medicine that encompasses many domains, including communication about medical diagnoses, informed consent, treatments, and prognoses, but also information gathering, relationship building, counseling, empathy, spiritual care, and grief support. Perhaps most importantly, communication involves active listening and compassionate presence to understand the family's values, goals, and unique needs. While high-quality communication should be an important element of every health care encounter, it is of particular importance at the end of a child's life when families are typically experiencing tremendously high levels of stress. Family-identified communication strategies that improve the end-of-life experience in the PICU include the following: empowering families to participate in decisions; limiting medical jargon; keeping families informed with complete, honest, and accurate information; not withholding poor prognosis; valuing the child as an individual; demonstrating compassionate, empathetic care of families; preparing families psychologically through the dying process; and displaying emotion and sharing family grief.¹³

The data for this study come from a previously completed multiinstitutional study that examined the current state of childhood death in the PICU setting. The first phase of the original study was a qualitative analysis of interviews of parents of children who had died in two study PICUs.¹⁴ The second phase of the original study collected data regarding the end-of-life care (including surveys of clinicians and medical record abstraction) for 169 children who died in five U.S. teaching hospital PICUs during a 15-month study period. A study describing the characteristics of children who die in PICUs in the United States was previously published.¹⁵ For this study, we examined the clinician surveys to understand their perspectives on communication with families at the end of a child's life. Specifically, we sought to understand the following: (1) where the majority of communication with the family occurs as this has not previously been reported, (2) if clinicians perceived barriers to caring for children in the PICU at the end of life as previously reported in adult ICUs, and (3) the potential impact of patient demographic characteristics, clinical factors, location of communication, and reported barriers to care on the quality of communication (QOC) as perceived by clinicians.

Patients and Methods

Sample

Data were collected between January 1, 2010, and March 31, 2011, at five PICUs in academic pediatric hospitals across the United States.¹⁵ In four institutions, data collection began on January 1, 2010, and concluded when 50 deaths had occurred or by December 31, 2010. In a fifth institution, data were collected on 24 deaths that occurred during an accrual period from October 1, 2010, through March 31, 2011. Within three days of every death, a site-based research coordinator identified five clinicians in specific roles, who had been most involved in the care of that child during the last three days of life: the bedside nurse, the primary nurse, the attending intensivist, the ICU fellow, and the primary psychosocial staff (social worker, psychologist, child life specialist, and chaplain). These clinicians were asked to complete self-administered surveys. Within each of the five sites, some clinicians completed surveys for multiple patients and most patients had surveys completed by multiple clinicians. Each survey represented the report of a particular clinician about the care provided for a particular patient.

Research ethics

Human subjects approval was granted by the Institutional Review Boards of all five participating hospitals and by Education Development Center, Inc., the nonprofit organization awarded funding to coordinate the project. Consent was obtained from all study participants before completion of the survey.

Measures

Demographic and clinical characteristics

Medical record abstraction was completed for each patient for age, gender, racial classification, PICU length of stay, illness trajectory (acute illness of a previously healthy child or acute or chronic illness progression of a child with chronic disease), nature of death (brain death, life support withdrawn or withheld, or unsuccessful cardiopulmonary resuscitation, CPR), and whether or not an active Do-Not-Resuscitate (DNR) order was in place. Demographic data were also collected for clinicians, including gender, race, years in practice, and years in critical care.

Location of communication

Clinicians were asked to estimate the frequency of family-clinician communication in which they participated or observed during the last three days of life in five locations: family conferences, rounds, bedside conversations, telephone conversations, and conversations in nonprivate places such as hallways and waiting areas. For each location, response options were none, less than half, about half, more than half, and most. To facilitate comparing the frequency with which communication occurred across the locations, the response for each location was recoded to a 2-category variable where more than half and most were coded as “majority” and other answers were coded as “nonmajority.”

Barriers to end-of-life care

Clinicians were asked about perceived barriers to helping patients and families in the PICU using an adapted version of the Barriers to Care survey tool.^16–18 This tool was previously validated to assess health care providers' perspectives on family-centered end-of-life care in the intensive care unit.

Quality of communication

Clinicians were asked to rate the quality of family-clinician communication in which they participated in, observed, or were aware of during the child's last three days of life using an adapted version of the Kalamazoo Essential Elements Communication Checklist.^19–21 Each clinician was asked to rate seven elements of communication with the family during the patient's last three days of life on a 5-point Likert scale (1 = poor and 5 = excellent). Seven communication elements included relationship building, information gathering, understanding the patient's and family's perspective, sharing information, reaching agreement, demonstrating empathy, and accurately communicating information. An overall QOC score was calculated as the mean of the individual items.

Statistical analysis

Descriptive statistics were used to summarize the demographic and clinical characteristics of the clinicians and patients, as well as clinician reports of the frequency of location of communication with families and the presence of barriers to care. Hierarchical linear regression modeling was used to account for clustering of different clinicians at different research sites. T-tests were used as preliminary univariate tests to help decide which variables to include in the multivariable model for estimating QOC score. Multivariable regression analysis was used to evaluate relationships between clinician-reported quality of family-clinician communication and location of communication and barriers to care, after controlling for research site, type of clinician, patient demographic characteristics (age, gender, and race), and clinical factors (length of stay, trajectory of illness, nature of death, and DNR order) with T-values and beta (β) coefficients of the final model. Statistical analysis was conducted using SAS 9.3 (SAS Institute, Cary, NC). A two-tailed p < 0.05 was considered significant.

Results

Clinician and patient demographics

Nine hundred forty-seven surveys were distributed to 430 clinicians. Five hundred sixty-five surveys were returned by 287 clinicians for 169 children who died. Of the 565 surveys, 38% were completed by nurses, 36% by physicians, and 26% by psychosocial staff (Table 1).

Table 1.

Survey Response Rates and Characteristics of Clinicians

Clinician characteristics	Total clinicians (n = 287a)	Bedside nurseb (n = 107)	Primary nurseb (n = 56)	Attendingc (n = 53)	Fellowc (n = 31)	Psychosocial staff (n = 67)
Survey data
Surveys distributed, no.	947	212	134	221	173	207
Surveys completed, no.	565	139	76	129	72	149
Clinician characteristics (%)
Female	77	96	89	29	59	86
Non-white	16	12	6	27	19	14
Years in practice
0–2 years	14	23	26	0	4	10
3–10 years	51	49	45	36	92	49
11+ years	35	28	30	64	4	41
Years in critical care
0–2 years	25	26	23	2	67	24
3–10 years	47	49	52	48	30	51
11+ years	28	26	25	50	4	25
EOL care training (% yes)	98	96	100	100	100	100

^aNumber of clinicians who completed one or more surveys.

^bNurses could serve as either the bedside or the primary nurse for any given patient. There were 24 nurses who completed a survey for one or more patients as a bedside nurse and another patient as a primary nurse. So the total number of nurses who completed surveys was 139 ( = 107 + 56–24).

^cNote that three clinicians served as a critical care fellow and then an attending physician during the accrual period. So the total number of physicians who completed surveys was 81 ( = 53 + 31–3).

EOL, end of life.

Median patient age was two years and just under half were female. Half (50%) of the patients were non-Hispanic Caucasians, 40% were non-white (15% African American, 17% Hispanic, 3% Asian, and 5% other), and 10% not documented. Median PICU length of stay was six days. Two-thirds of the patients had preexisting diseases, and one-third were healthy before PICU admission. A majority (71%) of patients died after withholding or withdrawing life-sustaining therapies, and half had an active DNR in place at the time of death. Basic patient characteristics are provided in Table 2 and a more detailed description of patient demographic and clinical characteristics is available in Burns et al.¹⁵

Table 2.

Patient Demographic and Clinical Characteristics

Patient characteristics	Number or
	median
Total patients	169a
Age
Median age, years (25th, 75th quartile)	2.0 (0, 9)
Less than 1 year, n (%)	58 (34)
21 years or older, n (%)	7 (4)
Gender, n (%)
Female	79 (47)
Male	90 (53)
Ethnicity and race, n (%)
Caucasian	84 (50)
African American	26 (15)
Hispanic	29 (17)
Asian	5 (3)
Other	9 (5)
Not documented	16 (10)
PICU length of stay before death
Median days in PICU (25th, 75th quartile)	5.7 (1.5, 20.3)
Less than seven days, n (%)	93 (55)
Seven days or more, n (%)	76 (45)
Illness trajectory, n (%)
Chronic or preexisting illness	103 (61)
New acute illness, previously healthy	66 (39)
Nature of death, n (%)
Withhold/withdraw life-sustaining treatments	120 (71)
Brain death	30 (18)
Failed cardiopulmonary resuscitation	19 (11)
DNR at time of death, n (%)
No	86 (51)
Yes	83 (49)

^aClinician surveys were completed on 169 of 192 patients who died during the study period. There were no significant differences in gender, age, race/ethnicity, or clinical characteristics between decedents who had surveys completed (n = 169, 88%) versus those decedents who did not (n = 23, 12%), both overall and within type of clinician.

PICU, pediatric intensive care unit; DNR, do not resuscitate.

Location of communication

Sixty-seven percent of surveys reported that the majority of communication during the last three days of life occurred at the bedside. The bedside was consistently reported as the most common location for communication regardless of patient demographic, clinical characteristics, or clinician type. Only 17% of surveys reported that the majority of communication for a given patient occurred during family conferences and only 16% during rounds. Unsurprisingly, relatively little communication was reported to have occurred over the phone or in nonprivate settings (Fig. 1).

FIG. 1.

Location of family-clinician communication during the last three days of life. Vertical bars represent percent of responses by type of clinician. Between 50% and 80% of surveys report that the majority of end-of-life communication occurs at the bedside, consistent across types of clinicians. Color image is available online.

Barriers to end-of-life care

The 10 examined barriers to caring for dying patients and their families were reported with frequencies of 2%–32% (Fig. 2). The most frequently reported barrier was “patient too sick to allow interaction with family” at 32%. Other barriers reported with frequencies of greater than 10% included “language differences,” “family had unrealistic expectations,” and “not enough staff, heavy patient load.”

FIG. 2.

Prevalence of barriers to care during the last three days of life. Horizontal bars represent the percentage of surveys that reported the presence of a given barrier. Eight of these barriers are associated with a lower QOC score when present. QOC, quality of communication.

Quality of communication

The overall QOC median score was 4.1 with an interquartile range between 3.7 and 4.7 (1 = poor and 5 = excellent). Based on regression analysis, the research site, type of clinician, patient age, gender, and clinical characteristics had no impact on QOC (Table 3). However, when the dying child was non-white, the QOC was rated lower (p = 0.03). Clinicians rated quality higher when the majority of communication took place during family conferences (p = 0.01). Eight of the 10 barriers had significant independently predictive negative effects on QOC (Table 3). Results of the multivariable analysis indicate a mean QOC of 4.4 on a 5-point Likert scale, given no barriers, and a lower score, given specific barriers, as indicated by corresponding beta coefficients in Table 3. For example, if language differences and lack of communication among the clinical staff were both reported, the estimated QOC score would be 3.85 points (4.4 – 0.24 – 0.31) on the 5-point scale.

Table 3.

Multivariable Model: Impact of Patient and Clinician Characteristics, Location of Communication, and Barriers to Care on Quality of Communication

Independent variables	t-value	p-value	β Coefficient (SE)
Intercept	36.18	<0.001	4.40 (0.12)
Patient's race
Non-white	−2.16	0.03	−0.13 (0.06)
Location of majority of communication
Family conferences	2.56	0.01	0.17 (0.07)
Rounds	1.69	0.09
Private at the bedside	1.24	0.22
Barriers to care
Too sick to allow interaction with family	−3.35	<0.001	−0.19 (0.06)
Language differences	−3.09	0.002	−0.24 (0.08)
Family had unrealistic expectations	1.49	0.14
Not enough staff, heavy patient load	−2.04	0.04	−0.16 (0.08)
Family was angry	−2.60	0.01	−0.22 (0.08)
Lack of communication among clinical staff	−2.94	0.004	−0.31 (0.10)
Conflict with others on the team	−2.15	0.03	−0.26 (0.12)
Family did not visit or call	−2.26	0.03	−0.28 (0.12)
Personal difficulty with family	−2.63	0.009	−0.45 (0.17)
Outside scope of practice	1.84	0.07

The multivariable model of the quality of communication included research site, type of clinician, patient characteristics (age, gender, race, PICU length of stay, illness trajectory, nature of death, and do-not-resuscitate order), location of communication (for the three most common locations), and barriers to care. Statistically significant findings are reported in bold.

The decrease in the QOC for non-white decedents and for those with language barriers raised the possibility of an interaction. Language barriers were more likely among non-whites (39.5%) compared to Caucasian (0.3%), but varied significantly by race/ethnicity among non-white patients. Language barriers were reported at 70% for Hispanic, 53% for Asian, 16% for other race, and 4% for African American. Small numbers precluded assessing this interaction in the full model. Based on t-tests comparing QOC by the presence or absence of a language barrier within non-white race/ethnicity categories, the only significant difference occurred for Hispanic families [mean = 4.4 (0.67) without language barrier and mean = 3.8 (0.65) with language barrier; p < 0.0001].

Discussion

Location of communication

To our knowledge, this is the first study to report on the location where PICU clinicians meet and talk with families. One of the most prominent findings to emerge from this study is that the majority of communication between clinicians and families occurs at the dying child's bedside. This finding was consistent across study site and type of clinician reporting. This finding has important implications for future research, given that very few studies examine communication that occurs in this setting.

Existing studies of ICU family-clinician communication focus almost exclusively on family care conferences^22–26 or ICU rounds.^27–29 Yet in actual PICU practice, rounds, bedside communication, and family care conferences are typically utilized in combination to support communication with families. Communication with families on bedside rounds typically focuses on the daily details of medical care. Bedside communication occurs when nurses care for their patients and interact with the families. Physicians talk to families at the bedside when examining patients, responding to medical needs, providing periodic updates, answering family questions, and engaging in decision making. Psychosocial staff often come to the patient's bedside to provide support for the family and the child. Bedside communication may be preferred by families who do not want to leave their dying child. Care conferences often encompass “big-picture” concerns, and can serve to coordinate care and decision making between myriad contributing team members and family members. The complex nature and process of communication in the PICU require that interventions to improve communication should focus on more than what occurs during family care conferences.

Future research should examine the different purposes and qualities of bedside communication, rounds, and care conference communication that support families at the end of life. Can we identify which families would benefit from certain styles of communication and how the approach might be customized to best meet the needs, values, and priorities of families? Can we identify elements of communication by location that we can target for improvement? What is the experience of the dying child when family-clinician conversations occur at the bedside? These questions remain unanswered at this time and should form the basis for future research and corresponding educational initiatives.

Quality of communication

Overall, QOC between clinicians and families of dying children is rated highly in the study PICUs by a wide range of clinicians. The relevance of this finding is significantly limited by the fact that previous work has clearly demonstrated that clinicians overestimate their own communication compared to objective third-party observers and patients.^30–35 However, our results do importantly identify factors that clinicians believe impact QOC. A better understanding of factors that negatively affect QOC can help to focus education and quality improvement efforts.

Key variables that impact QOC

Patient race

It is concerning to find that patient and family non-white race continue to be negatively associated with QOC. Recent studies in the PICU corroborating this finding have reported that non-white race is associated with a lower likelihood of discussing limitation or withdrawal of life support,³⁶ associated with increased depression and PTSD after the death of a child,¹⁰ and more likely to experience failed resuscitation as the mode of death.³⁷ Our data suggest that a coinciding language barrier may contribute to lower QOC among Hispanic families, but it does not appear to be a significant influencing factor among families of other races. Future studies examining QOC and race should include a more detailed examination of language barriers, including the influence of interpreters. Despite ongoing attention to health care disparities, differences remain. Ongoing research, education, and quality improvement efforts are of utmost importance to eliminate this disparity.

Location of communication

Clinicians rated QOC higher when the majority of communication took place during family conferences (p = 0.01). The reasons behind this finding are unclear and we caution against an interpretation that more communication at the end of life should occur in the setting of a family care conference. It may be that clinicians perceived QOC was better, given that family care conferences often have an agenda, are more organized, and are more easily observed by a variety of PICU clinicians. Preferred location of communication would be very important to validate with families in future studies.

Clinician-reported barriers to care

Clinicians reported that barriers to caring for the patient and family were present 2%–32% of the time. Of the 10 barriers examined, 8 of them negatively impacted QOC.

One barrier stands out with a relatively high prevalence of nearly one-third of cases: “patient too sick to allow interaction with family.” One possible explanation is that the stress of not being able to interact with their child made it more difficult for the family to engage in quality communication with the clinicians. An alternate explanation is that clinicians were occupied with the acute care of the dying child and this prevented communication with the family. One speculation is that this barrier was present at times of attempted CPR; however, in our study, this barrier was not associated with nature of death. Developing a better understanding of this barrier will be important to determining effective interventions. Most circumstances in the PICU should allow for quality care of the family and quality communication by a skilled clinician.³⁸

Other barriers that were reported with high frequencies included “language differences,” “family had unrealistic expectations,” “not enough staff, heavy patient load,” “family was angry,” and “lack of communication among clinical staff” (Fig. 2). These findings are supported by previous studies.^39–41 While these barriers are familiar to clinicians, they are potentially modifiable, and understanding their importance can help to prioritize them for interventions.

We cannot make assumptions about the nature of interventions to address these barriers based on these data. A more granular assessment of how clinicians define and interpret these barriers is needed. Nonetheless, identification that these barriers are present and negatively impact QOC is a step forward to ultimately developing educational and clinical interventions to improve quality of PICU communication.

Study limitations

Clearly, the most important assessment of QOC at the end of a child's life is that of the family. The addition of data from families regarding their experiences of communication would be an important adjunct to this study. Despite considerable effort to recruit families to this study, the response rate was low (∼20%). Future studies focused on understanding the family experience and perspective will be critical additions to these data. Nonetheless, the findings from this study describing the current state of clinician assessment of QOC at the end of life in the PICU help us to direct future efforts to improve communication.

Another important limitation of this study is that in the intervening time since data collection, PICU cultural changes have occurred regarding family-clinician communication, family-centered care, and even changes to physical spaces within PICUs for family communication. It would be important to reassess these findings in future studies.

Conclusions

Communication is the cornerstone of high-quality care of dying children and their families in the PICU. We identified several important targets for additional research and education as we continue to strive to improve communication at the end of life. Further research is needed to better understand the nature of bedside communication and how different locations of communication and types of communication strategies can be customized to meet individual families' needs and priorities. In addition, further investigation is needed to explore strategies to lessen the impact of barriers to care on communication at the end of life. Ongoing research must continue to increase understanding of how and why race influences communication and interactions with families. Finally, future research is required to better understand the patient and family experience of communication at the end of life.

Funding Information

Funded by National Institute of Nursing Research R01 NR09298-01A1.

Author Disclosure Statement

No competing financial interests exist for any of the authors.