Intervenciones para la comunicación interpersonal acerca de los cuidados paliativos terminales entre los profesionales sanitarios y las personas afectadas
Resumen
Antecedentes
La comunicación acerca del final de la vida (FdV) y los cuidados terminales es de vital importancia para proporcionar una atención de calidad cuando las personas afrontan la muerte. Esta comunicación suele ser compleja e implica a muchas personas (pacientes, familiares, cuidadores, profesionales sanitarios). No se sabe cuál es la mejor manera de comunicarse con las personas en el periodo próximo a la muerte, pero es una cuestión importante para la calidad de los cuidados paliativos terminales en todo el mundo. Esta revisión aporta evidencia sobre la comunicación interpersonal (entre las personas y los profesionales sanitarios) en el último año de vida, centrándose en las intervenciones para mejorar la comunicación interpersonal y los desenlaces de los pacientes, los familiares y los cuidadores.
Objetivos
Evaluar los efectos de las intervenciones diseñadas para mejorar la comunicación verbal interpersonal al FdV sobre los cuidados paliativos terminales entre los profesionales sanitarios y las personas afectadas.
Métodos de búsqueda
Se realizaron búsquedas en CENTRAL, MEDLINE, Embase, PsycINFO y CINAHL desde el inicio hasta julio de 2018, sin restricciones de idioma ni fecha. Se estableció contacto con los autores de los estudios incluidos y con expertos y se realizaron búsquedas en las listas de referencias para identificar documentos pertinentes. En septiembre de 2019 se realizaron búsquedas en fuentes de literatura gris, resúmenes de congresos y registros de ensayos clínicos. Las búsquedas en las bases de datos se volvieron a realizar en junio de 2021 y los estudios potencialmente relevantes se mencionaron como pendientes de clasificación o en curso.
Criterios de selección
Esta revisión evaluó los efectos de intervenciones, evaluadas en ensayos aleatorizados y cuasialeatorizados, destinadas a mejorar la comunicación interpersonal sobre los cuidados paliativos terminales entre los pacientes con pronóstico de morir en menos de 12 meses, sus familiares y cuidadores, y los profesionales sanitarios implicados en su atención. Fueron elegibles los pacientes de cualquier edad, en cualquier contexto o ámbito asistencial (p. ej., lesión mortal aguda, enfermedad crónica), y todos los profesionales sanitarios implicados en su atención. Todas las intervenciones de comunicación fueron elegibles, siempre que incluyeran interacción/es interpersonal/es entre los pacientes y los familiares o cuidadores y los profesionales sanitarios. Las intervenciones podrían ser simples o complejas, con uno o más objetivos de comunicación (p. ej., informar, capacitar, comprometer, apoyar). Se buscaron los efectos sobre los desenlaces de los pacientes, la familia y los cuidadores, los profesionales sanitarios y los sistemas de salud, incluidos los efectos adversos (no deseados).
Para asegurar que esta revisión se mantuviera centrada en la comunicación interpersonal en los últimos 12 meses de vida, se excluyeron los estudios que abordaron decisiones específicas, compartidas o no, y las herramientas implicadas en dicha toma de decisiones. También se excluyeron los estudios centrados en las voluntades anticipadas que informaron sobre la aceptación o el cumplimiento de las voluntades anticipadas como desenlace principal. Por último, se excluyeron los estudios de formación en habilidades comunicativas para profesionales sanitarios, a menos que los desenlaces de los pacientes se presentaran como desenlaces principales.
Obtención y análisis de los datos
Se utilizó la metodología estándar de Cochrane, que incluye la selección de estudios, la extracción de los datos y la evaluación de la calidad de los estudios incluidos por dos autores de la revisión.
Resultados principales
Se incluyeron ocho ensayos. Todos evaluaron los efectos de la intervención en comparación con la atención habitual. La certeza de la evidencia fue baja o muy baja. La certeza de todos desenlaces se disminuyó por las medidas indirectas en función del objetivo de la revisión, y en el caso de muchos desenlaces, se disminuyó por imprecisión o inconsistencia. Por lo general, la certeza no se disminuyó por limitaciones metodológicas.
A continuación se resumen las conclusiones de la revisión.
Conocimiento y comprensión (cuatro estudios, evidencia de certeza baja; un estudio sin datos utilizables): las intervenciones para mejorar la comunicación (p. ej., lista de preguntas, con o sin formación del paciente y del médico) podrían tener poco o ningún efecto sobre el conocimiento de la enfermedad y el pronóstico, o las necesidades y preferencias de información, aunque los estudios fueron pequeños y las medidas utilizadas variaron entre los ensayos.
Evaluación de la comunicación (seis estudios que midieron varios componentes [calidad de la comunicación, cuánto se centró en el paciente, preferencias de la participación, relación médico‐paciente, satisfacción con la consulta], la mayoría evidencia de certeza baja): en todos los componentes podría haber efectos mínimos o nulos de las intervenciones para mejorar la comunicación al FdV, y hay incertidumbre sobre los efectos de intervenciones como una hoja de comentarios específica para el paciente sobre la calidad de la comunicación.
Debates sobre el FdV o la atención paliativa terminal (seis estudios que midieron desenlaces seleccionados, evidencia de certeza baja o muy baja): una intervención de reunión familiar podría aumentar la duración de las conversaciones sobre el FdV en el ámbito de la unidad de cuidados intensivos (UCI), mientras que el uso de una guía de conversación estructurada sobre las enfermedades graves podría dar lugar a conversaciones más tempranas sobre el FdV y la atención paliativa terminal (cada intervención evaluada por un estudio). No están claros los efectos sobre la frecuencia de las conversaciones y la formulación de preguntas en las consultas, y podría haber poco o ningún efecto sobre el contenido de la comunicación en las consultas.
Desenlaces adversos o efectos no deseados (evidencia limitada): no hay evidencia suficiente para determinar si hay desenlaces adversos asociados con las intervenciones de comunicación (p. ej., lista de preguntas, reunión familiar, debates estructurados) al FdV y la atención paliativa terminal. La ansiedad del paciente y del cuidador se comunicó en tres estudios, pero se consideró confusa. No se informó sobre otras consecuencias no deseadas, ni un empeoramiento de los desenlaces deseados.
Calidad de vida del paciente/cuidador (cuatro estudios, evidencia de certeza baja; dos sin datos utilizables): las intervenciones para mejorar la comunicación podrían tener poco o ningún efecto sobre la calidad de vida.
Desenlaces de los profesionales sanitarios (tres estudios, evidencia de certeza baja; dos sin datos utilizables): las intervenciones para mejorar la comunicación podrían tener poco o ningún efecto sobre los desenlaces de los profesionales sanitarios (satisfacción con la comunicación durante la consulta; un estudio); se desconocen los efectos sobre otros desenlaces (conocimientos, preparación para la comunicación).
Impactos en los sistemas de salud: las intervenciones de comunicación (p. ej., conversaciones estructuradas sobre el FdV) podrían tener poco o ningún efecto en las calificaciones de los cuidadores o de los médicos sobre la calidad de la atención paliativa terminal (satisfacción con la atención, control de los síntomas, evaluación de la comodidad, calidad de la atención) (tres estudios, evidencia de certeza baja) o sobre la atención y enfermedad autoevaluada por los pacientes, o el número de metas de atención cumplidas (un estudio, evidencia de certeza baja). Las intervenciones de comunicación (p. ej., la lista de preguntas sola o con formación en habilidades comunicativas dirigida por personal de enfermería) podrían aumentar ligeramente la duración media de la consulta (dos estudios), pero no están claras otras repercusiones en los servicios sanitarios (p. ej., los ingresos hospitalarios).
Conclusiones de los autores
Los resultados de esta revisión no son concluyentes para la práctica. Los estudios de investigación futuros podrían contribuir de manera significativa a completar las carencias en las poblaciones que aún no se han estudiado en los ensayos; y a desarrollar criterios de valoración sensibles con los que evaluar mejor los efectos de la comunicación en el abanico de personas implicadas en los episodios de comunicación al FdV. Los estudios de investigación con métodos mixtos y cualitativos podrían contribuir de forma útil a comprender mejor la compleja interacción entre las diferentes partes implicadas en la comunicación, y a informar sobre el desarrollo de intervenciones más eficaces y medidas de desenlace adecuadas. El diseño conjunto de dichas intervenciones y desenlaces, con la participación de todo el abanico de personas afectadas por la comunicación y los cuidados paliativos terminales, debería ser un principio fundamental para los estudios de investigación futuros en esta área.
PICO
Resumen en términos sencillos
¿Cómo se puede mejorar la comunicación acerca del final de la vida y los cuidados paliativos en los últimos 12 meses de vida?
Mensajes clave
No se encontró suficiente evidencia de buena calidad para poder decir qué formas de comunicación acerca del final de la vida (FdV) son las mejores para las personas implicadas. Un estudio sobre una intervención de reunión familiar determinó que las intervenciones de comunicación podrían aumentar la duración de las conversaciones al FdV entre las familias y los profesionales sanitarios en algunas situaciones, y otro determinó que una intervención que utilizó una guía de conversación estructurada podría dar lugar a conversaciones más tempranas entre pacientes, cuidadores y profesionales sanitarios sobre el FdV y los cuidados paliativos terminales. No se encontró evidencia de efectos perjudiciales o negativos de las intervenciones de comunicación, ni se sabe con certeza los efectos sobre desenlaces como el conocimiento o la calidad de los cuidados paliativos terminales.
¿Por qué es importante la comunicación al final de la vida?
Cuando alguien se encuentra en el último año de su vida, es importante que reciba una atención de calidad (consulte las referencias de ACSQHC 2015 y 2015b para obtener más información sobre los cuidados paliativos terminales). La comunicación acerca del FdV es una parte fundamental de esta atención. Ayuda a los pacientes y a sus familiares y cuidadores a entender lo que está sucediendo, a saber qué esperar y cuáles son las opciones, a hacer preguntas y recibir apoyo, y a participar en las decisiones y la planificación tanto como lo deseen. La comunicación acerca del FdV no siempre se hace bien, lo cual puede tener efectos negativos. Entender cómo mejorar esa comunicación entre las diferentes personas que participan en los cuidados paliativos terminales (pacientes, familiares, cuidadores, profesionales sanitarios) es importante para ayudar a garantizar que las personas reciban la mejor atención posible en el tiempo que precede a la muerte.
¿Qué se quería averiguar?
Se quería averiguar qué formas de comunicación con los pacientes y los cuidadores podrían ser las mejores para mejorar los conocimientos de las personas sobre el FdV (p. ej., qué esperar, opciones de tratamiento).
La opinión de las personas acerca de la comunicación (p. ej., satisfacción, calidad de la comunicación, grado de participación y deseo de participar en las consultas). Conversaciones sobre el final de la vida (p. ej., con qué frecuencia se producen y cuándo).
También se quería averiguar si las intervenciones de comunicación podían aumentar los efectos no deseados o perjudiciales, como el miedo o la angustia.
¿Qué se hizo?
Se buscaron estudios que analizaran las intervenciones de comunicación en comparación con la atención habitual (la atención que se proporciona de forma rutinaria o como la forma estándar de tratar a las personas), o que compararan un tipo de comunicación (p. ej., proporcionar información) con otro (p. ej., proporcionar información junto con apoyo), en personas de todas las edades y que se esperaba que murieran en el transcurso de 12 meses. Se resumieron los resultados de los estudios incluidos y la confianza en la evidencia se calificó en función de factores como el tamaño del estudio, los métodos de estudio y las personas estudiadas por los ensayos.
Para asegurar que esta revisión se mantuviera centrada en la comunicación interpersonal en los últimos 12 meses de vida, se excluyeron los estudios que abordaron decisiones específicas, compartidas o no, y las herramientas implicadas en dicha toma de decisiones. También se excluyeron los estudios centrados en las voluntades anticipadas que informaron sobre la aceptación o el cumplimiento de las voluntades anticipadas como desenlace principal. Por último, se excluyeron los estudios de formación en habilidades comunicativas para profesionales sanitarios, a menos que los desenlaces de los pacientes se presentaran como desenlaces principales.
¿Qué se encontró?
Se encontraron ocho estudios que compararon los efectos de las intervenciones de comunicación para personas al FdV con la atención habitual. Las intervenciones fueron variadas y abarcaron desde estrategias sencillas, como una lista para ayudar a los pacientes y cuidadores a hacer preguntas en las consultas, hasta complejas intervenciones de conversación estructurada para fomentar la participación de los pacientes y los cuidadores en las conversaciones sobre el FdV y la atención que deseaban recibir.
Se encontró que una intervención de reunión familiar podría aumentar la duración de las conversaciones sobre el FdV en algunas situaciones, y una guía de conversación estructurada sobre enfermedades graves podría llevar a conversaciones más tempranas entre pacientes, cuidadores y profesionales sanitarios sobre el FdV y los cuidados paliativos terminales.
También se observó que las intervenciones de comunicación podrían tener poco efecto sobre los conocimientos, sobre lo que las personas piensan de la comunicación (p. ej., la calidad de la comunicación, el grado de participación en la conversación que les gustaría tener) o sobre los desenlaces, como el número de preguntas que hacen los pacientes en las consultas con sus médicos. No se encontró evidencia de efectos perjudiciales o negativos de las intervenciones, pero los estudios eran en su mayoría pequeños y no estaban diseñados principalmente para identificarlos.
También podría haber poco efecto en los otros desenlaces que se analizaron, como la calidad de vida, la calidad de los cuidados paliativos terminales o el número de metas de atención alcanzadas. En otros casos, no hay certeza porque hubo poca o ninguna evidencia disponible (p. ej., los desenlaces de los profesionales sanitarios como los conocimientos o la confianza a la hora de comunicarse, o el uso de los servicios de salud, p. ej., los ingresos hospitalarios).
¿Cuáles son las limitaciones de la evidencia?
Se tiene muy poca confianza en la evidencia: los estudios incluidos sólo analizaron la comunicación en adultos mayores de países de ingresos altos, a pesar de que la revisión buscó evidencia en todo el ciclo vital y sin importar el país ni el entorno. Además, los estudios incluidos a menudo estudiaron un número reducido de personas.
¿Cuál es el grado de actualización de esta evidencia?
La evidencia está actualizada hasta julio de 2018.
Authors' conclusions
Summary of findings
| Communication intervention compared with usual care for end of life care | |||
|---|---|---|---|
| Patient or population: people approaching the end of life (within 12 months), their family members and/or carers Settings: any (residential care, hospital (inpatient and outpatient units) and community‐based clinics, palliative care services) Intervention: interventions to improve communication about EoL and/or EoL care Comparison: usual care | |||
| Outcomes | Intervention effects | Number of participants | Certainty of the evidence |
| Patient, family and/or carer outcomes | |||
| Knowledge and understanding
Variable scales: information (amount and type) needs/preferences; discordance between patient and physician survival and curability estimates
Timing: immediately to 1 month post‐consultation | Overall, interventions to improve communication may have little or no effect on measures of knowledge of illness and prognosis, or information needs and preferences
In 1 study (303 participants), discordant estimates of 2‐year survival between patients and doctors (intervention 59% versus usual care 62%) and curability (intervention 39% versus usual care 44%) were similar between groups (Epstein 2017). Another (79 participants) reported similar proportions of patients had their preferences for amount of information met or exceeded across intervention and usual care groups, but that type of information was met or exceeded more often in the intervention group (93% versus 80%) (Walczak 2017). 1 final study (170 participants) reported no differences in patients’ unmet information needs overall (Clayton 2007) | 552 (3 studies)c | ⊕⊕⊝⊝
|
| Evaluation of the communication: different constructs (perceptions of communication quality; patient‐centredness of communication; involvement preferences; doctor‐patient relationship measures)
Timing: immediately post‐consultation to 18 weeks post‐consultation | Across constructs (patient‐centredness, involvement preferences, doctor‐patient relationship, satisfaction with consultation), there may be minimal or no effects of interventions to improve communication about EoL and EoL care (Agar 2017; Bernacki 2019; Clayton 2007; Epstein 2017; Walczak 2017), and uncertaintye about effects on quality of communication (Au 2012)
| 6 studiesf | ⊕⊕⊝⊝ |
| Discussions of EoL/EoL care: discussion timing and length
EMR review post‐death
Timing: at time of family conference (intervention) in ICU; post‐death
| The intervention may lead to longer and earlier discussions of EoL and EoL care, compared with usual care, but each result is based on a single study
1 study (108 participants) reported comparative data: median family conference intervention duration was 30 minutes (IQR 19 to 45 minutes) versus usual care (median 20 minutes, IQR 15 to 30 minutes) (Lautrette 2007)
1 study (376 participants) reported that the first documented Serious Illness Conversation happened earlier among intervention group participants (median 143 days prior to death (IQR 71 to 325) than usual care (71 days, IQR 33 to 166) (Bernacki 2019) | 484 (2 studies) | ⊕⊕⊝⊝ |
| Discussions of EoL/EoL care: discussion occurrence
EMR review post‐death; coding of consultations; self‐reported occurrence
Timing: immediately, 1 or 2 weeks post‐consultation; after death | Overall, we are uncertain about the effects of interventions to improve discussions about EoL care
2 studies indicated that the intervention increased the occurrence of EoL discussions, compared with usual care (RR 1.96, 95% CI 1.61 to 2.39; 2 trials, 537 participants); the others indicated little or no effect of the intervention on mean total numbers of patient questions in consultations (MD 1.58, 95% CI ‐1.82 to 4.98; 2 trials, 249 participants) | 786 (4 studies) | ⊕⊝⊝⊝
|
| Adverse (unintended) outcomes
| There is insufficient evidence to determine whether adverse (unintended) outcomes are associated with communication interventions. Patient and/or carer anxiety was reported (3 studies), but was judged as confounded, and no other unintended consequences, or worsening of desired outcomes, were reported | ‐ | ‐ |
| CI: confidence interval; EMR: electronic medical record; EoL: end of life; ICU: intensive care unit; IQR: interquartile range; MD: mean difference; RR: risk ratio | |||
| GRADE Working Group grades of evidence | |||
| aDowngraded (‐1) due to inconsistency (different outcome measures and concepts assessed across studies). | |||
Background
Description of the condition
Discussion about end of life (EoL) between health practitioners and affected people can be a confronting experience for all parties involved. According to the Australian Medical Association, "Death, dying and bereavement are all an integral part of life; however, reflecting on, and discussing death, can be profoundly confronting and difficult. Open and frank discussion of death and dying including EoL care options, approach to futile treatment, caring and bereavement should be encouraged within the profession and in the wider community" (AMA 2014). EoL and EoL care can be defined in many different ways. For this review, we have adopted the following definitions, developed as part of a national (Australian) consensus statement on end of life care.
-
End of life: "the period when a patient is living with, and impaired by, a fatal condition, even if the trajectory is ambiguous or unknown. This period may be years in the case of patients with chronic or malignant disease, or very brief in the case of patients who suffer acute and unexpected illnesses or events, such as sepsis, stroke or trauma" (ACSQHC 2015, page 33).
-
End of life care: "includes physical, spiritual and psychosocial assessment, and care and treatment delivered by health professionals and ancillary staff. It also includes support of families and carers, and care of the patient’s body after their death. People are 'approaching the end of life' when they are likely to die within the next 12 months. This includes people whose death is imminent (expected within a few hours or days) and those with:
-
advanced, progressive, incurable conditions;
-
general frailty and co‐existing conditions that mean that they are expected to die within 12 months;
-
existing conditions, if they are at risk of dying from a sudden acute crisis in their condition; and
-
life‐threatening acute conditions caused by sudden catastrophic events" (ACSQHC 2015, page 33).
-
Taken together, the above definitions show that the EoL period may be one of prognostic uncertainty and highly variable in duration. This review acknowledges this uncertainty and the difficulties associated with defining this period. When the selection criteria for this review were developed, the definition given by the above statement (i.e. people are approaching EoL when they are expected to die within 12 months) was the most recent available definition for Australian audiences and so was adopted as a working definition to define the scope of the review.
People involved in communication with health practitioners about EoL and EoL care may include the person at EoL and the family or carers of that person (Hjelmfors 2020; Wolfe 2020). Each of these people may have an important role in discussions about EoL care. For the purpose of this review, we needed to define these different people in a way that is not ambiguous, given the multiplicity of terms that are used in different health systems for all parties. Further, although the term 'patient' is not always suitable for someone who may often not be in a patient role, we needed to distinguish the person at EoL from that person's family member or carer. We therefore define affected people as follows.
-
Patient: identified as "the primary recipient of care" (ACSQHC 2015, page 34). In many health systems and countries, terms other than 'patient' are preferred. However, in this review we use this term to distinguish clearly between people who are approaching the end of their life, or dying (and to whom discussions about prognosis, treatment, and care relate directly), and people to whom these discussions relate indirectly (i.e. discussions about EoL and EoL care related to a family member or person in whose care they are involved).
-
Family: this review takes the broadest possible view of family members, considered to represent "those who are closest to the patient in knowledge, care and affection. This may include the biological family, the family of acquisition (related by marriage or contract), and the family and friends of choice" (ACSQHC 2015, page 33). It also includes First Nations definitions of family within the wider culture, such as those encompassed by the concept of kinship care (Palliative Care Australia 2016).
-
Carer: "a person who provides personal care, support and assistance to another individual who needs it because they have a disability, medical condition (including a terminal or chronic illness) or mental illness, or they are frail and aged. An individual is not a carer merely because they are a spouse, de facto partner, parent, child, other relative or guardian of an individual, or live with an individual who requires care" (ACSQHC 2015, page 32).
The focus of this review is interpersonal interactions occurring between patients, family, carers, and health practitioners at the EoL.
EoL discussions are often placed within the context of palliative care. The "WHO [World Health Organization] identified that, globally, palliative care needs are very high, with an estimated 20 million people needing end‐of‐life care each year" (AIHW 2014, page 2). This enormous demand exists across countries and healthcare systems (World Wide Palliative Care Alliance 2014), yet palliative care is only one of the contexts in which good communication about EoL care is essential.
Internationally, a large body of research has documented difficulties in EoL communication between healthcare professionals and people affected by EoL (i.e. patients, their families, and carers) (Anderson 2019; Clayton 2007a; Fawole 2012; Fujimori 2020; IoM 2014; NICE 2017; Walczak 2016). These difficulties include failure to communicate adequately with the person who is dying about his or her prognosis (Barnes 2006; Brighton 2016; Fawole 2012; Gott 2009; NICE 2017), or to provide understandable information on what the future holds, and decisions that the person and family members and carers may need to make (Alsakson 2012; Anselm 2005; Barnes 2012; Gutierrez 2012; NICE 2019; Selman 2007). It is also documented that patients receiving EoL care, or those closest to them, may not be given the opportunity to ask questions or to check their understanding of information that has been provided (Alsakson 2012; Clayton 2007a; Gutierrez 2012; Hjelmfors 2020). People often have misunderstandings about their prognosis and goals of treatment in the EoL period (Anderson 2019; Clayton 2007a; ; Gattellari 1999; Thode 2020; Weeks 1998). Misunderstandings may also arise from conflicting information given by multiple practitioners involved in the patient's care. Additionally, the patient and family members or carers may have their own questions about EoL care but may be unaware of how or whom they should approach to find answers to these questions (Alsakson 2012; Anselm 2005; Gutierrez 2012; NICE 2017).
Communication problems have significant potential to negatively impact the person who is dying and family members or carers. Communication problems may contribute to loss of trust in health practitioners (Clayton 2007a; NICE 2017), poorer quality of life and satisfaction, psychological harms, and avoidable distress (Chochinov 2000; Fawole 2012; Fujimori 2020; NICE 2017; Schofield 2003; Selman 2007; Wright 2008). These negative outcomes reflect poorly on the ability of existing healthcare systems to effectively deliver patient‐centred, responsive care during EoL (Anderson 2019; IoM 2014; NICE 2017; NICE 2019). In comparison, high‐quality communication about EoL is associated with improved quality of life and less aggressive approaches to treatment, as well as better outcomes for carers related to bereavement (Brighton 2016; Detering 2010; Heyland 2009; Wright 2008; Zhang 2009).
Communicating effectively about EoL is a difficult and complex task that is further complicated by uncertainty about the trajectory of the last stages of a person's life and the associated prognosis (Barnes 2012; Brighton 2016; Fawole 2012). Good communication remains the primary means of preparing the patient and affected people for the last months, weeks, and days of life (Anderson 2019; Fawole 2012). This review focuses on general communication between health practitioners and people affected by EoL ‐ not specifically on communication involving use of specific tools to achieve structured decision‐making (such as communication or discussion in which participants use a highly structured checklist to develop an advance care plan). Such specific, structured decision‐making approaches were therefore excluded. With focus on general communication, this review is able to evaluate the evidence on interventions intended to improve communication about EoL care among health professionals and patients and their family members and carers. Previous research has shown that discussion of prognosis and EoL is important to people who are dying and to their families (Brighton 2016; Fujimori 2020; NICE 2019; Steinhauser 2000; Walczak 2016; Wenrich 2001; Wolfe 2020). It is clear that for people to be able to articulate what they would like, and to participate in decisions about their care in the last stages of their life or the life of someone close to them, they must be adequately informed (Clayton 2007a; NICE 2019). A recent study seeking to develop quality indicators for EoL communication and decision‐making confirmed that these discussions are also important to health professionals and the systems in which EoL care is delivered (Sinuff 2015). The highest‐rated quality indicator overall was related directly to whether discussions about prognosis and the likelihood that the patient is approaching the end of life had actually been undertaken (Sinuff 2015). This review therefore seeks to evaluate how communication about EoL and EoL care might be better undertaken, and to assess the impact of verbal communication on the various people most directly involved.
Description of the intervention
Communication interventions can be broadly defined as "a purposeful, planned and formalised strategy associated with a diverse range of intentions or aims, including to inform, educate, communicate with, support, skill, change behaviour, engage and seek participation of people" (Hill 2011, page 30). This review follows this broad view and considers a communication intervention as a planned interaction provided by health practitioners to communicate with people about EoL and provision of EoL care. Although these interventions may take many forms and may reflect different purposes, to be eligible for this review interventions must have included direct interpersonal (verbal) communication between health practitioners and the patient and the patient's family members and/or carers. Specifically, these interventions could have taken the form of facilitating or improving EoL care discussions targeting a broad range of continuum of care, ranging from rapidly evolving situations to early preparatory stages of what may be a protracted period of terminal care. The review included any EoL communication intervention that involved a patient who was likely to die within 12 months (ACSQHC 2015, page 2; NICE 2017, page 7). Communication interventions must have been primarily interpersonal (verbal) in nature and preferably delivered in‐person, although if necessary they may have included the following channels for communication: in‐person, telephone, videoconferencing, remote video links, and internet‐enabled verbal discussions. Other non‐verbal forms of communication, such as written information, may also have been included as part of the intervention, and while data were collected on these approaches they were not the primary focus of the review.
The intervention may have focused on one or more of the following elements of EoL or EoL care: knowledge of what might happen around the disease and what a possible disease trajectory might be for the patient (prognosis); understanding of the possibilities for treatment, pain management, symptom management, and treatment or care to relieve suffering; preferences for care or treatment or both, including wishes regarding the location of living until dying; needs or concerns related to supportive, spiritual, cultural, or palliative care; needs or concerns related to the role of the family or carer, including support for family members/carers; needs or concerns associated with administrative paperwork, formal documentation, dying or the choice for assisted dying (for jurisdictions where relevant); and death. The intervention may have been tailored towards an individual or a small group, as long as the group includes patients and their family members or carers.
We considered the full range of EoL communication interventions identified as eligible for this review, and we anticipated that their dispersion and application across studies might vary. The needs and circumstances of the people involved were also expected be complex and highly varied. Accordingly, the elements of EoL and EoL care discussed were expected to be tailored to specific EoL contexts. EoL discussions are not limited to a specific healthcare setting, so it was important that this review was inclusive of EoL communication interventions applied irrespective of national, geographical, cultural, social, wealth, and healthcare access boundaries. Such diverse EoL experiences could be related to gender, ethnicity, race, religion, culture, refugee status, indigenous peoples, gender diversity, disability, socioeconomic status, education, poverty, and populations in low‐ and middle‐income countries (Welch 2010). For this reason, the review considered inequality and inequity issues as they relate to EoL communication interventions (Welch 2010).
To ensure this review's focus was maintained on interpersonal communication in the last 12 months of life, we excluded studies that addressed specific decisions, shared or otherwise, and the tools involved in such decision‐making. We also excluded studies focused on advance care planning (ACP) reporting ACP uptake or completion as the primary outcome. Finally, we excluded studies of communication skills training for health professionals unless patient outcomes were reported as primary outcomes.
How the intervention might work
Interventions to improve EoL verbal communication aim to provide more effective general communication between practitioners and the people directly affected by EoL and EoL care. Previous reviews have confirmed the highly complex and varied scope of EoL experiences, and support the need for the study intervention to be fully described and to include EoL context, details of what the intervention entails, and related primary patient outcomes (Fawole 2012; Walczak 2016).
We have described the content of the EoL communication intervention above. Practitioners could use a variety of modalities to deliver the intervention and to guide or influence the discussion about EoL or EoL care. Examples could include prompts for patients to promote or guide discussions about EoL care (Clayton 2007b; Fujimori 2020; Hjelmfors 2020; Sansoni 2014; Walczak 2017); web‐based collaboration tools to facilitate communication between practitioners and people affected by EoL (Voruganti 2017; Walczak 2016); nurse‐led discussions about EoL care (Sulmasy 2017); or EoL family meetings (Agar 2017; Bradford 2021; Walczak 2016). Outcomes chosen to measure effects of the interventions could reflect changes in the level of communication occurring (e.g. increasing the frequency or length or both of discussions between practitioners and patients and affected people), improved structure of the communication taking place (e.g. providing prompts to assist patients, family members, and carers to ensure that key questions are raised with practitioners, thereby improving knowledge and understanding about EoL care), or specific outcomes related to patient's/affected people's EoL care experiences and their experiences of the communication around EoL.
Why it is important to do this review
General EoL communication guidelines are already available. For example, in 2007, Medical Journal of Australia published a supplement titled 'Clinical practice guidelines for communicating prognosis and end of life issues with adults in the advance stages in a life limiting illness, and their caregivers' (Clayton 2007a). More recently published EoL guidelines related to paediatric patients and young people include the 'End of life planning series' (Together for Short Lives 2012), along with 'Difficult Conversations' (Together for Short Lives 2015). EoL care standards and quality markers and measures of EoL care related to communication are also available (ACSQHC 2015; NICE 2017). A more recent exploratory study conducted with paediatric practitioners confirmed that evidence‐based interdisciplinary interventions are needed to support general EoL discussions (Henderson 2017). A systematic review of communication quality improvement interventions for patients with advanced and serious illness completed in 2012 confirmed that better descriptions of communication interventions were needed for assessment of impact on the outcomes being researched (Fawole 2012). Although general guidelines on communication are available, they do not necessarily address or draw on rigorous research evidence related to the effectiveness of specific EoL communication interventions.
A systematic review and meta‐analysis undertaken by Oczkowski in 2016 examined communication tools for EoL decision‐making in ambulatory care settings (Oczkowski 2016). The Oczkowski review was focused on EoL decision‐making and advance care planning and concluded that use of structured communication tools should be the preferred approach to EoL decision‐making conversations (Oczkowski 2016). Another recent systematic review of studies of mixed designs (Thode 2020) assessed the role of communication tools such as decision aids for people considering life‐prolonging treatments. It similarly concluded that prompt lists and decision aids may be useful in communicating with patients about options for treatment, but was based on a small number of studies in a population that is not directly relevant to this current review. Two completed Cochrane Reviews ('Advance care planning for haemodialysis patients' (Lim 2016) and 'Advance care planning for adults with heart failure' (Nishikawa 2020) also have an indirect link with this review. The current review includes discussions on the topic of advance care planning, but only when these conversations are taking place in the last 12 months of life, and only when uptake of advance care planning (ACP) or advance directives (AD) is not the primary goal of the study.
Several other Cochrane Reviews, for example 'End‐of‐life care pathways for improving outcomes in caring for the dying' (Chan 2016); 'Hospital at home: home based end of life care' (Shepperd 2021); and 'Communication skills training for healthcare professionals working with people who have cancer' (Moore 2018), have addressed issues related to EoL, but they have not addressed the interventions to improve communication, with a distinct focus on patient outcomes, explored in this current review.
Studies of ACP or AD that did not meet these criteria were therefore deliberately excluded as they focused on the outcomes of the process (i.e. ACP completion) rather patient outcomes (this review's focus). Additionally, these strategies are commonly not closely related in time to the end of life, with many elderly people now asked to undertake ACP in preparation for death that may be years or even decades in the future. This variability in degree of temporal linkage to EoL, as well as heavy reliance on checklists and structured tools common with ACP, also led to the exclusion of these studies. As indicated above, studies focusing on specific decisions using structured tools (e.g. decision aids) were excluded in order to ensure the review maintained a focus on patient outcomes and how these were influenced by communication.
Communication skills training for health professionals was also excluded from this review, unless patient outcomes were reported as primary outcomes. This decision aligned with the reasoning above, as interventions to prepare professionals to communicate typically focus on evaluating the effectiveness of such strategies to improve clinician skills (how, and how well, clinicians communicate) ‐ a step influencing but preceding the communication encounter with patients, and typically reflected in a lesser focus on patient outcomes. One of the review's main underpinning principles was that interventions involved interpersonal interaction between health practitioner(s) and the patient, family, and/or carers in order that the focus on patients be maintained.
It is worth noting that had we included studies with a focus on structured decision‐making tools like those underpinning many approaches to ACP, or those on communication skills training, this review would have quickly become unfeasible and run to inclusion of potentially hundreds of trials ‐ as this represents a very substantial literature. Such a review however, would have a far more dispersed scope, and it would have been very difficult to untangle the effects of interpersonal communication for the people involved at EoL within this larger collective body of research.
Previous reviews of the literature have considered EoL communication interventions. Barnes 2012 undertook a critical review of the literature to explore patient‐professional communication about EoL issues in life‐limiting conditions. These review authors found limited evidence regarding successful interventions to improve discussions with patients about EoL care. Additionally, communication topics are often embedded in more specific EoL research. Walczak 2016 completed an important systematic review of evidence for EoL communication interventions. This review identified 45 studies through a search of the literature conducted in 2014, and review authors concluded that "Overall, greater use of validated measures, commonality of outcomes between studies and meta‐analyses allowing more concrete statements about the efficacy of end‐of‐life communication interventions are vital to the advancement of the field" (Walczak 2016, page 13). Bradford 2021 completed a systematic review regarding family meetings in paediatric palliative care, finding there was little guidance about how meetings should be organised or conducted, or when these should occur. Overall, the literature confirms that there is general agreement that EoL communication and interventions to improve such communication are important for providing quality care for patients and other people affected by EoL.
To inform how EoL communication can be improved in future practice, one must gain an understanding of the effectiveness of communication interventions in the EoL context, and the impact these interventions can have on measurable outcomes for patients, families, and carers. The findings of this review should prove important in this endeavour. Improved and more effective communication between health practitioners and people affected by EoL has the potential to help practitioners address gaps in care and to improve poor outcomes such as distress and lower quality of life associated with poor communication (Brighton 2016). This will provide a foundation where patients and others affected by EoL events can participate in shared decisions about treatment and care.
Objectives
To assess the effects of interventions designed to improve verbal interpersonal communication about end of life (EoL) care between health practitioners and people affected by EoL.
Methods
Criteria for considering studies for this review
Types of studies
We included randomised and cluster‐randomised controlled trials (RCTs) and quasi‐RCTs evaluating the effects of interventions intended to enhance communication between health practitioners and patients and families or carers about end of life (EoL) care. We expected to find a limited number of RCTs on this topic and therefore planned to include quasi‐RCTs (defined as trials attempting, but not achieving, random allocation of participants).
Types of participants
We included the following participants.
-
Patients with a life‐limiting illness who were expected to die within 12 months (ACSQHC 2015).
-
Patients with cancer, end‐stage pulmonary disease, end‐stage cardiac failure, end‐stage renal failure, motor neuron disease, or other chronic conditions (e.g. dementia), as reported in the study.
-
Patients with a life‐threatening acute condition caused by sudden catastrophic events (ACSQHC 2015).
-
Vulnerable groups of patients with a life‐limiting illness, as reported in the study. For example, patients could be in a third world setting in which EoL is not explicitly defined. In such cases, researchers may use terms such as 'dying' and 'death', which can be used to identify the study as relevant.
-
Patients of any age from birth who met one of the criteria listed above.
We also included family or carers of a patient with a life‐limiting illness, as defined by the study. We defined family as "biological, family of acquisition (related by marriage or contract) and the family and friends of choice" (ACSQHC 2015, page 33). We defined a carer as "a person providing personal care, support and assistance for the patient with a life‐limiting illness" (ACSQHC 2015, page 32).
We did not exclude studies based on the setting of the communication or the person delivering the communication, although the communication must have involved a health practitioner. We defined health practitioners for inclusion in this review as follows.
-
Healthcare professionals may include doctors, nurses, midwives, allied health practitioners, social workers, and government healthcare workers.
-
The professional population could be identified as the healthcare team, the interdisciplinary team, the interprofessional team, or a group of healthcare providers, as reported in the study.
-
We included lay health workers, who are not health practitioners as such but who are educated/trained to deliver the intervention (e.g. may be applicable in resource‐poor/low‐ and middle‐income country settings or within a specific cultural context to promote cultural safety).
-
We included other community providers or volunteers, as reported in the study.
Types of interventions
We included any interventions provided to promote or improve interpersonal communication between health practitioners and people affected by EoL care versus usual care. We also included comparisons of one form of communication intervention versus another.
The communication may have focused on any aspect of EoL or EoL care, including the following.
-
Knowledge of what might happen around the disease and what a possible disease trajectory might be for the patient (prognosis).
-
Understanding of the possibilities for treatment, pain management, symptom management, and treatment or care to relieve suffering.
-
Preferences for care or treatment or both (e.g. resuscitation, feeding), including wishes regarding the location of living until dying.
-
Needs or concerns related to supportive, spiritual, cultural, or palliative care.
-
Needs or concerns related to the role of the family or carer, including support for family members/carers.
-
Needs or concerns associated with administrative paperwork, formal documentation, and dying or the choice for assisted dying (for jurisdictions where relevant).
The intervention must have involved interpersonal interaction between health practitioner(s) and the patient, family, and/or carers. We included videoconferencing, remote video links, or internet‐enabled discussions only if the parties involved could not be located physically together (e.g. in the case of patients living in rural, remote, or underserved areas).
The communication intervention might have included one or more of the following aims: to inform or educate, support, skill, engage, or seek the participation of patients and their families and carers in a communication episode with professionals around EoL care. Interventions could be simple or complex; we included interventions as long as the effects of the communication element of any complex intervention could be isolated by inclusion of an appropriate comparison group.
We excluded the following studies.
-
Studies focusing on specific decisions ‐ shared or otherwise. This review focused on general communication between health practitioners and patients and their family members and carers. Such communication may be viewed as a necessary and fundamental precursor to more specific decisions about treatment and other choices, which may often involve highly structured or specific communication tools (as described above).
-
Studies focusing on development or completion of an advance care planning (ACP) or advance directives (AD) for which uptake or completion is the primary outcome.
-
Studies assessing the effects of public education (e.g. on ACP), or of general individual education (e.g. about ACP, or about how to speak up).
-
Studies focusing on case conferencing for specific decision‐making needs, or case conferencing about choice of residence (e.g. discharging patient to a nursing home or to a palliative care service).
-
Studies focusing on communication skills training for health professionals (unless patient outcomes were reported as primary outcomes).
-
Studies involving health practitioner communication with a group of people, unless that group comprised the patient, family members, and/or carers.
Types of outcome measures
We collected data on a range of primary and secondary outcomes.
Primary outcomes
Patient, family, and/or carer (affected persons) outcomes
-
Knowledge and understanding about what might happen (prognosis), or what to do, or options.
-
Evaluation of the communication ‐ positive constructs (e.g. satisfaction, calmness or confidence about ability to manage the future).
-
Evaluation of the communication ‐ negative constructs (e.g. fear, anxiety, distress).
-
Discussions of EoL care/EoL (e.g. frequency, length, type, participants).
Adverse outcomes
-
Any adverse outcomes or harms identified in the included studies.
-
These might have included any negative effects on the primary outcomes listed above.
-
Secondary outcomes
-
Health practitioner knowledge and understanding of patient/family/carer knowledge, wishes, or preferences.
-
Health practitioner evaluation of his or her communication performance, the overall communication encounter, or self‐confidence or preparedness to communicate.
-
Patient/family member/carer quality of life.
Health systems impacts relevant to the impacts of communication
-
Costs of subsequent care.
-
Hospital admissions and re‐admissions (e.g. hospital bed days, intensive care unit (ICU) admissions).
-
Quality of EoL care (family/carer rated, practitioner rated).
-
Ratings of concordance with patient preferences for EoL care.
We did not exclude studies that were otherwise eligible based on the outcomes reported, except for the situation described above, in which the intervention focused on ACP/AD and the primary outcome sought was uptake or completion.
Main outcomes for the summary of findings tables
We reported the following outcomes.
Patient, family, and/or carer (affected persons) outcomes
-
Knowledge and understanding about what might happen (prognosis), what to do, or options.
-
Evaluation of the communication ‐ positive constructs (e.g. satisfaction, calmness or confidence about ability to manage the future, preparedness to plan for the future).
-
Evaluation of the communication ‐ negative constructs (e.g. fear, anxiety, distress).
Adverse events
-
These were reported as any negative changes in the above outcomes associated with the intervention.
We reported findings for each of the primary outcomes in the summary of findings tables.
If multiple outcomes were reported in a given outcome category, we collected information on all relevant outcomes. However, if the same outcome had been assessed by two or more outcome measures in the same trial, we planned for two review authors to:
-
select the primary outcome measure identified by the publication authors;
-
select the one specified in the sample size calculation when no primary outcome measure was identified; and
-
rank effect estimates (i.e. list them in order from largest to smallest) and select the median effect estimate if no sample size calculations were reported.
-
When an even number of outcome measures was reported, we planned to select the outcome measure whose effect estimate was ranked n/2, where n was the number of outcome measures.
-
Search methods for identification of studies
Electronic searches
We searched the following electronic databases in July 2018, all from inception.
-
Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library (to 27 July 2018).
-
MEDLINE (OvidSP) (1946 to 27 July 2018).
-
Embase (OvidSP) (1947 to 27 July 2018).
-
PsycINFO (OvidSP) (1806 to 27 July 2018).
-
Cumulative Index to Nursing and Allied Health Literature (CINAHL) (EBSCOHost) (1937 to 27 July 2018).
Search strategies for all major databases are presented in Appendix 1 to Appendix 2.
There were no language or date restrictions.
We applied the Cochrane RCT Classifier to the database search results. The Classifier assigned a probability (from 0 to 100) of being a randomised trial to each citation retrieved. Those citations with a Classifier scores of nine or less were excluded from dual reviewer screening but were screened by a single reviewer (titles and abstracts) as part of a check of the accuracy of the Classifier and to ensure that no studies were misclassified and wrongly excluded from the search outputs.
Database searches were re‐run in June 2021. Studies potentially meeting the selection criteria are listed in studies Awaiting classification.
Searching other resources
We contacted experts in the field and authors of included studies for advice as to other relevant studies and searched reference lists of relevant studies. We searched (September 2019) relevant grey literature sources (ProQuest Dissertations & Theses, British Library Electronic Theses Online Service (EThOS)), conference proceedings (European Association for Palliative Care (EAPC), American Society of Clinical Oncology (ASCO), World Congress of Psycho‐oncology), and clinical trials registries (US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov and World Health Organization International Clinical Trials Registry Platform (WHO ICTRP)) to identify relevant trials.
Data collection and analysis
Selection of studies
Two review authors independently screened all titles and abstracts identified through searches to determine which met the inclusion criteria. We retrieved in full text any papers identified as potentially relevant by at least one review author. Two review authors independently screened full‐text articles for inclusion or exclusion and resolved discrepancies by discussion and by consultation with a third review author if necessary to reach consensus. We listed all potentially relevant papers excluded from the review at this stage as excluded studies and provided reasons for exclusion in the 'Characteristics of excluded studies' table. We also provided citation details and any available information about ongoing studies and collated and reported details of duplicate publications, so that each study (rather than each report) is the unit of interest in the review. We report the screening and selection process in an adapted PRISMA flow chart (Liberati 2009); see Figure 1.
Study flow diagram.
Screening was performed by at least two review authors working independently, except those citations classified with a score of nine or less by the RCT Classifier, which were screened by a single review author. Citations from conference proceedings and trials registries were also screened by a single review author, who consulted with a second review author on any potentially relevant studies.
Data extraction and management
Two review authors extracted data from included studies. Ratings of risk of bias were made independently by two review authors, otherwise data were extracted by one review author and checked for accuracy by a second. We resolved any discrepancies by discussion until consensus was reached, or through consultation with a third review author when necessary. We developed and piloted a data extraction form using the Cochrane Consumers and Communication Group (CCCG) Data Extraction Template (available at cccrg.cochrane.org/author-resources)). Data extracted included the following: study details (aim of intervention, study design, description of the intervention and comparison group, outcomes, and data). One review author entered all extracted data into Review Manager 5 (Review Manager 2020), and a second review author working independently checked the data for accuracy against the data extraction sheets.
Assessment of risk of bias in included studies
We assessed and reported on the methodological risk of bias of included studies in accordance with the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), as well as the guidelines of the Cochrane Consumers and Communication Group (Ryan 2013), which recommend explicit reporting of the following individual elements for RCTs: random sequence generation, allocation sequence concealment, blinding of participants and personnel, blinding of outcome assessment, completeness of outcome data, selective outcome reporting, and other sources of bias (baseline imbalances). We considered blinding separately for different outcomes when appropriate (e.g. blinding may have the potential to differently affect subjective versus objective outcome measures). We judged each item as being at high, low, or unclear risk of bias as set out in the criteria provided by Higgins 2011, and provided a quote from the study report or a justification for our judgement or both for each item in the risk of bias table.
We judged studies to be at highest risk of bias if they scored as at high or unclear risk of bias for either the sequence generation or the allocation concealment domain, based on growing empirical evidence that these factors are particularly important potential sources of bias (Higgins 2011).
In all cases, two review authors independently assessed the risk of bias of included studies and resolved disagreements by discussion to reach consensus. We contacted study authors for additional information about the included studies, or for clarification of study methods as required. We incorporated results of the risk of bias assessment into the review through standard tables and systematic narrative description and commentary about each of the elements, leading to an overall assessment of the risk of bias of included studies and a judgement about the internal validity of results of the review.
We planned to assess and report quasi‐RCTs as being at high risk of bias on the random sequence generation item of the risk of bias tool, but none were identified. For cluster‐RCTs, we also assessed and reported the risk of bias associated for an additional domain: selective recruitment of cluster participants.
Measures of treatment effect
For dichotomous outcomes, we analysed data based on the number of events and the number of people assessed in the intervention and comparison groups. We used these numbers to calculate the risk ratio (RR) and the 95% confidence interval (CI) where possible. For continuous measures, we analysed data based on the mean, the standard deviation (SD), and the number of people assessed for both intervention and comparison groups to calculate the mean difference (MD) and 95% CI. If the MD was reported without individual group data, we had planned to use this information to report the study results. If more than one study measured the same outcome using different tools, we calculated the standardised mean difference (SMD) and the 95% CI using the inverse variance method in Review Manager 5.
Unit of analysis issues
We checked cluster‐RCTs for unit of analysis errors. Levels of allocation and analysis were different in all four cluster‐RCTs, but all appropriately adjusted for clustering in their analyses.
If errors and sufficient information were available, we planned to re‐analyse the data using the appropriate unit of analysis, by taking account of the intracluster correlation (ICC). We planned to obtain estimates of the ICC by contacting authors of included studies, or impute them using estimates from external sources. Where it was not possible to obtain sufficient information to re‐analyse the data, we planned to report effect estimates, annotated with 'unit of analysis error'.
Dealing with missing data
We attempted to contact authors of all included studies to obtain missing data (participant, outcome, or summary data), or to clarify details of the trial's methods or conduct or both. All but two teams of authors were successfully contacted and provided additional information about their trial. We planned to analyse participant data based on an intention‐to‐treat basis, however we analysed the data as reported. We reported on the levels of loss to follow‐up and assessed this as a source of potential bias.
For missing outcome or summary data, we planned to impute missing data when possible, report any assumptions in the review, and investigate the effects of imputed data on pooled effect estimates through sensitivity analysis. We were unable to conduct these analyses due to the small number of studies contributing data for all outcomes.
Assessment of heterogeneity
When we considered studies similar enough (based on consideration of populations, interventions, or other factors) to allow pooling of data using meta‐analysis, we assessed the degree of heterogeneity by visually inspecting forest plots and by examining the Chi² test for heterogeneity. We quantified heterogeneity by using the I² statistic. We considered an I² value of 50% or more to represent substantial levels of heterogeneity, but we interpreted this value in the light of size and direction of effects and strength of the evidence for heterogeneity, based on the P value derived from the Chi² test (Higgins 2011).
We planned, when we detected substantial clinical, methodological, or statistical heterogeneity across included studies, not to report pooled results from meta‐analysis but instead to use a descriptive approach to data synthesis. However, where we judged studies to be similar enough clinically and methodologically to justify statistical pooling, and data were available, but heterogeneity was high, we reported the pooled result irrespective of high variability and accounted for this in our GRADE ratings of evidence certainty.
At protocol stage we planned to attempt to explore possible clinical or methodological reasons for variation across studies descriptively synthesised by grouping studies that were similar in terms of populations, intervention features, methodological features, or other factors to explore differences in intervention effects. However, numbers of studies contributing data to any one outcome were small and did not allow this type of analysis to go ahead.
Assessment of reporting biases
We assessed reporting bias qualitatively based on the characteristics of included studies (e.g. if only small studies that indicated positive findings were identified for inclusion), or where information that we obtained upon contacting experts and study authors suggested that there were relevant unpublished studies.
If we had identified sufficient studies (at least 10) for inclusion in the review, we planned to construct a funnel plot to investigate small‐study effects, which may indicate the presence of publication bias. In such instances, we planned to formally test for funnel plot asymmetry, after choosing the test based on advice provided in Higgins 2011, and bearing in mind when interpreting study results that there may be several reasons for funnel plot asymmetry.
Data synthesis
We decided whether to meta‐analyse data based on whether interventions in the included trials were similar enough in terms of participants, settings, comparisons, and outcome measures to ensure meaningful conclusions from a statistically pooled result. Owing to anticipated variability in populations and interventions, and possibly other factors, we used a random‐effects model for meta‐analysis.
Where we were unable to pool the data statistically using meta‐analysis, we prepared a descriptive synthesis of results. We presented data, organised by major outcome categories, and subcategories where applicable, in tables and in text. We had planned to explore the main comparisons of the review (intervention versus usual care; one intervention form versus another) within data categories but only the first comparison was assessed by included studies.
For each outcome/data category, we drew together results of meta‐analysis or descriptive synthesis or both to provide an overall synthesis of the effects of the intervention for each outcome category and subcategory.
Subgroup analysis and investigation of heterogeneity
We did not anticipate including enough studies with quantitative data to warrant subgroup analyses, but planned to attempt to explore potential effects of the following factors through systematic grouping of studies and synthesis when possible.
-
Type of EoL care: groupings might include palliative care, acute (emergency) care, and others. The rationale for considering effects separately in such (or similar) groupings is that communication needs, opportunities to communicate, and information and decisions needed are likely very different across such different types of EoL care.
-
Type or aim or both of intervention: groupings might include those to inform and educate, those to support communication, and those to promote communication or decision‐making skills. The rationale for separately considering these groupings is that interventions with different purposes have different underlying mechanisms of action.
Too few studies contributing data to any outcome were included in the review to enable the above analyses to proceed.
Sensitivity analysis
As anticipated, we did not include enough studies in any one pooled analysis to justify conducting sensitivity analyses. However, in future if we identify sufficient studies, we will consider removing those rated as having the highest risk of bias from the analysis and examining effects on the pooled effect estimates.
Ensuring relevance to decisions in health care
One of the co‐authors (Josephine Bothroyd (JB)) is a consumer representative for the Healthcare Consumers' Association of the Australian Capital Territory. She had input to the protocol at all stages.
We planned to consult more widely about the consumer perspective with consumer groups, industry, and/or government agencies. However, given the inconclusive findings of the review we did not perform these wider consultations at this stage. This may be an avenue to explore in future updates of the review.
A consumer provided feedback on the protocol and the review as part of standard CCCG editorial processes.
Summary of findings and assessment of the certainty of the evidence
We prepared a summary of findings table to present results for the main outcomes, based on the methods described in Chapter 11 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011). We presented the results of analysis for the major comparison of the review, for each of the major primary outcomes, including potential harms, as outlined in the Types of outcome measures section. We used the GRADE system to rank the certainty of evidence (GRADEpro GDT; Schünemann 2011). Two or more review authors independently assessed outcomes against the GRADE criteria, with discussion to reach consensus on final ratings of certainty.
Results
Description of studies
Results of the search
Database searches identified 21,170 records for screening. The RCT Classifier was used to assess these references, with 11,069 references remaining for screening with a 90% or higher likelihood of being true randomised controlled trials (RCTs). These references were screened, together with another 245 citations identified from searches of grey literature and other sources (e.g. author contact). After de‐duplicating references, two review authors independently screened 10,970 abstracts. Of these, 10,672 were excluded, with 298 papers screened in full text. In total 251 of these papers were excluded, four ongoing studies were identified, two are awaiting classification, and eight studies (reported in 40 papers) were included in the review; see Figure 1 for PRISMA chart.
One review author (Rebecca Ryan (RR)) screened the remaining 10,101 records excluded by the RCT Classifier as being of lower likelihood of being RCTs. No studies were identified for inclusion in the review from this secondary screening.
Included studies
Trial and participant features
Full details of the included studies are given in Characteristics of included studies tables.
We included eight trials, see Table a below. In four, participants were individually randomised to one of two arms, while in the remaining four, clusters of participants were randomised to one of two (three trials) or three (one trial) arms. All cluster trials appropriately adjusted for clustering in their analyses. See Additional Table 1 for participant numbers for each trial.
| Participant numbers | ||||||||
|---|---|---|---|---|---|---|---|---|
| Eligible for inclusion | Nursing homes: 111 eligible Participants: 148 UC 171 intervention | 1292 1173 mailed introductory letter | Clinicians: 133 recruited Patients: 9182 screened UC (8530 ineligible) 9395 screened intervention (8842 ineligible) | 196 | Physicians: 38 eligible Patients: 453 eligible | 132 | 214 | 363 |
| Excluded | Nursing homes: 91 (55 did not meet criteria; 36 declined) 14 UC 11 intervention | 30 screened out (did not meet criteria) 21 declined participation prior to screening | Clinicians: 6 pilot clinicians ineligible Patients: 8987 UC 9211 intervention | 22 | Patients: 137 excluded (38 ineligible, 99 refused) | ‐ | ‐ | ‐ |
| Refused to take part | ‐ | 645 (did not wish to participate) 101 (did not keep appointment) | Clinicians: 36 Patients: unclear | 18 | 99 | 6 | 104 | 253 |
| Randomised to intervention group(s) | 10 homes, 160 participants 156 received allocated intervention | 194 | Clinicians: 48 (20 clusters) Patients: 184 (20 clusters) | 92 | Physicians: 19 Patients: 139 Carers: 105 | 63 | 58 | 61 |
| Randomised to control (usual care) group | 10 homes, 134 participants 130 received allocation intervention | 182 | Clinicians: 43 (21 clusters) Patients: 195 (19 clusters) | 82 | Physicians: 19 Patients: 142 Carers: 99 | 63 | 52 | 49 |
| Excluded post‐randomisation (for each group; with reasons if relevant) | UC 66 Intervention 89 All excluded because did not die during the study period | ‐ | 96 (intervention) Death; lost to follow‐up; declined further surveys; completed 2 years 118 (UC) Death; lost to follow‐up; declined further surveys; completed 2 years Intervention: 184 (20 clusters) enrolled: 134 analysed (18 clusters); (50 total: no family/friend response, no baseline survey, withdrew) 34 (13 clusters) analysed: 74 died, 11 lost to follow‐up; 17 declined further surveys; 32 completed 2 years UC: 195 (19 clusters) enrolled: 144 analysed (17 clusters); (51 total: no family/friend response, no baseline survey, withdrew) 26 patients (13 clusters) analysed: 77 died. 12 lost to follow‐up, 21 declined further surveys; 34 completed 2 years | ‐ | ‐ | ‐ | ‐ | ‐ |
| Withdrawn (for each group; with reasons if relevant) | UC 4 Intervention 4 (died before intervention) | Control 27 (14.5%) 3 patient‐clinician relationship changed 6 refuse to continue 9 not contactable 8 no target visit 1 too ill/deceased Intervention 43 (22.2%) 4 patient‐clinician relationship changed 15 refuse to continue 8 not contactable 10 no target visit 3 too ill/deceased 3 other | Intervention 12 Control 8 | None | Intervention 6 Control 5 | Intervention 11 Control 7 | ‐ | ‐ |
| Lost to follow‐up (for each group; with reasons) | ‐ | ‐ | Intervention: 11, no reason given Control 12, no reason given | Intervention: 2, 1 mistakenly seen by a junior physician who was not participating in the study, 1 due to mechanical failure of tape recorder Control: 2, 1 mistakenly seen by a junior physician who was not participating in the study 1 due to mechanical failure of tape recorder | Intervention: 3 died Control: 1 died 1 lost to follow‐up | Intervention: did not answer telephone n = 4; experiencing severe emotional distress n = 5; refused interview n = 2 Control: did not answer telephone n = 3; experiencing severe emotional distress n = 1; refused interview n = 2; patient still alive n = 1 | Total numbers reported at each time point N = 96/110 completed 3‐month measures N = 90/110 completed 6‐month measures Plus an additional 3 where complete data were not available NB also for several outcomes data for slightly lower numbers in total are available i.e. table 3 – ranges from n = 65 to n = 81. Not clear what happened to these missing measures | Intervention: 21 (34%) Unclear; data collection hampered by declining health, attrition high (higher in intervention group) but no systematic reasons for differential dropout identified. Reasons for loss not reported specifically Control: 9 (18%) Reasons for loss not reported specifically |
| Included in the analysis (for each group, for each outcome) | Intervention 67 Control 64 These numbers were analysed throughout, with losses for particular scales/assessments noted where applicable | Intervention 194 Control 182 | Intervention 38 patients (13 clusters) Control 26 patients (13 clusters) | Intervention 90 Control 80 | Intervention 19 physicians, 130 patients Control 19 physicians, 135 patients | Intervention 52 (83%) Control 56 (89%) | 6‐month data (longest time point) Intervention 47 Control 40 | Intervention 39 Control 40 |
| Assessment of attrition bias for RoB ratings | Assessed as low risk Large proportion of data (participants randomised) missing. However, these were comparable for the 2 study groups, and was due to participants not dying during the study period (outcomes assessed for this study were focused on those around death) Withdrawal rates for other reasons were low and comparable across groups | Assessed as unclear risk Withdrawal 15% to 22% respectively control and intervention arms Reasons for withdrawal/dropout were reasonably comparable except that more (15 versus 6) refused to continue participation in the intervention group Authors report no differences in baseline characteristics regarding whether patients completed the study or were lost to follow‐up ITT analysis was used; effect of imputed data on results was examined in analysis models with authors reporting similar results where imputed and non‐imputed data were used in analysis | Assessed as unclear risk Patient participation rates and numbers analysable were low but comparable between arms | Assessed as low risk Low levels of loss to follow‐up 4/174; balanced across groups (n = 2 each), with comparable reasons | Assessed as low risk "Fewer than 3% of follow‐up questionnaires were missing" (page 95) Data seem otherwise complete for outcomes reported in main paper and in supplement 3 | Assessed as low risk Loss to follow‐up and withdrawals were acceptably low and comparable across study groups: 52/63 (83%) completed interviews at 90 days intervention group, 56/63 (89%) in control group Reasons for withdrawal/loss were similar across groups (not answering telephone, refused interview), although higher rates of severe emotional distress in intervention group (n = 5) than control group (n = 1) | Assessed as unclear risk Missing data were reported However, numbers were lower for some outcomes (such as ratings of care management) where n = 65 (numbers fairly comparable between the 2 groups) | Assessed as high risk Attrition was high (31/110 (28%) lost to follow‐up), possibly largely explained by declining health of participants (patients) Authors state that ITT analysis was used (according to group assignment) but dropout rates may be problematic |
| Additional notes | The study was underpowered for the primary outcome because fewer participants died during the study period than predicted Predicted sample size required recruitment of 272 participants (assuming 10% dropout rate) (17 per site) | ‐ | Sample size calculated as 200 evaluable patients per arm for required power, assuming 6% dropout | ‐ | ‐ | ‐ | ‐ | Trial is underpowered (sample size calculated at 140; 110 recruited; 79 completed) to detect differences between groups |
ITT: intention to treat; RoB: risk of bias; UC: usual care.
All studies were conducted in high‐income countries (four USA, three Australia, one France), and all in urban settings mostly associated with larger hospitals, clinics, or residential care facilities. All participants were older patients where the mean age was 60 years or more, despite inclusion criteria being wide (aged 18 years and older) for most studies, and even in the single study conducted in an intensive care unit (ICU) setting (Lautrette 2007). Most studies included minority ethnic groups to a small degree, and most explicitly excluded people not fluent in the majority language (English or French) and those with cognitive impairment.
Gender composition varied across trials: in the two dementia trials most patients were female (approximately 60% to 80%), whereas gender in chronic obstructive pulmonary disease (COPD) patients in Veteran's Affairs centres was almost exclusively male (96% or more). Other trials fell between these extremes. Only one trial reported surrogates' characteristics, with 70% or more of surrogate decision makers for ICU patients being female.
In two studies, patient participants had advanced dementia and the interventions targeted carers/ family members as surrogate decision makers, as did the study of patients in ICU where surrogate decision makers (usually family members) received the intervention. In remaining studies participants had diagnoses of advanced cancer and other advanced progressive diseases, and both the patient and carer or family member (and sometimes also professionals) were stated targets of the intervention.
Table a: Major trial and participant features
| Study | Country, setting
| Diagnosis of person at end of life (EoL); selection criteria | Demographics | Intervention target |
|---|---|---|---|---|
|
Cluster‐RCT 2 arms
| Australia
Residential care facilities in major cities | Advanced dementia; selection criteria identified people with average survival < 6 months
Carers no information
| Age (years): intervention mean 84.7 (standard deviation (SD) 7.9), usual care (UC) 85.8 (SD 8.2)
Gender: intervention 61% female, UC 58% female | Residents with dementia Family member/ friend involved in decisions on patient's behalf |
|
Cluster‐RCT 2 arms
| USA
Outpatient clinics, Veterans' Affairs centres | COPD Clinician primarily responsible for COPD care (primary care and chest clinics)
Excluded: cognitive dysfunction, language barriers or severe psychiatric disorders
| Age (years): patients mean 69.4 both arms
Gender: patients: intervention 97.9% male, UC 96.2% male clinicians: intervention 50% male, UC 44% male | Patients with COPD Clinicians Surrogates |
|
Cluster‐RCT 3 arms (2 control; data from 2 arms available)
| USA
Hospital clinics | Advanced, incurable cancer; life expectancy < 12 months and identified surrogates (family member/ friend)
Clinicians: oncology physicians, nurse practitioners, physician assistants caring for patients (advanced incurable cancer and life expectancy < 12 months) at least 1‐half day per week
Excluded: cognitive impairment, unable to speak English or identify a surrogate. Clinicians participating in concurrent studies, or working with both intervention and control arms | Age (years): patients: intervention 61.8, UC 62.1
Gender: patients: intervention 53.7% female, UC 52.8% female clinicians: intervention 62.5% female, UC 51.2% female | Patients with cancer Clinicians Surrogates |
|
RCT 2 arms
| Australia
Palliative care services | Advanced, progressive, life‐limiting conditions Carer (spouse, partner, family member, friend)
Clinicians: palliative care physicians who endorsed question prompt list (QPL) use
| Age (years): intervention: mean 65.5 (SD 12.6), UC 64.6 (SD 14.1)
Gender: intervention 39% female, UC 40% female | Patients Carer
|
|
Cluster‐RCT 2 arms
| USA
Community‐based cancer clinics, academic medical centres, community hospitals
| Advanced cancer; mean life expectancy 9 to 12 months Carers (family member, partner, friend, other involved in health care) Oncologists
Excluded: inpatients, those in hospice, unable to understand spoken English or provide written informed consent, patient/carer without decisional capacity | Age (years): patients: mean 64.4 carers: not reported physicians: mean 44
Gender: patients: 55% female carers: not reported physicians: 29% female | Patients Carers Oncologists
73% of patients nominated a carer
|
|
RCT 2 arms
| France
Intensive care (medical and surgical) units
| ICU (acute respiratory failure, coma, shock, acute renal failure, cardiac arrest; expected to die within days)
Excluded: patients < 18 years, surrogates without sufficient French for telephone interview
| Age (years): patients: intervention median 68, UC 74 surrogates: intervention median 54, UC 54
Gender: patients: intervention 41% female, UC 48% female surrogates: intervention 77% female, UC 70% female | Surrogate decision makers (primarily family members): 40% spouses, 48% children
|
|
RCT 2 arms
| USA
Large nursing facility in major city (New York)
| Advanced dementia Family members/surrogates (patient’s primary contact)
| Age (years): intervention: mean 59.6 (SD 12.3), UC mean 58.9 (SD 11.9)
Gender: intervention: 78.7% female, UC 80% female | Surrogates
|
|
RCT 2 arms
| Australia
Hospital‐affiliated cancer treatment centres | Advanced, incurable cancer; oncologist‐assessed life expectancy 2 to 12 months
Informal primary carers > 18 years participated if nominated by patient
Excluded: non‐English speaking, those with cognitive impairment or significant psychological morbidity | Age (years): intervention: mean 63.8, UC: mean 65.6
Gender: 34.5% female
| Patients, with or without primary informal carers
|
Intervention and comparison features
All studies assessed the comparison intervention versus usual care; see Additional Table 2. The single three‐arm study (Bernacki 2019) included two control arms (one usual care, one control), with data available only for the usual care arm. Usual care showed some variability in levels of contact and support received, but most often participants received routine consultations or family conferences, with one (Reinhardt 2014) including social contacts via telephone in addition to usual care, in order to control for the effects of structured follow‐up calls in the intervention group.
| STUDY ID |
| |||||||
|---|---|---|---|---|---|---|---|---|
| Intervention(s) aim and components | To improve EoL care
Facilitated CC + patient‐centred palliative care training | To improve communication about preferences for EoL care
1‐page patient‐specific feedback form based on patient’s self‐reported responses | To evaluate the SICG
SICG + training, including support for response documentation and patient/family materials | To evaluate QPL effects on patient/carers’ EoL care topic discussions in consultations
QPL for patients and caregivers | To improve patient‐centred communication between physicians and patients/carers
Complex patient‐centred communication training (2 components: physician, patient) | To improve communication between family and ICU staff and support family decision‐making
Proactive communication family conference plus bereavement information leaflet | To provide information and support to surrogates of patients with advanced dementia
Face‐to‐face structured conversation, telephone follow‐up | To increase patients’ EoL care discussions and cue oncologists to endorse QPL and question asking
Nurse‐led CSP (QPL, booster and verbal/written cueing of oncologists pre‐consultation) |
| Comparator (usual care) | Usual care (no additional education, training, support) | Usual care (no patient‐specific feedback) | Usual care | Usual care (routine consultation with PC physician) | Usual care (oncologist met with research assistant but received no training) | Usual care (routine family EoL conference); informs family of treatment limitations and that death is imminent; led by senior physician, with at least 1 family member | Usual care plus social contact by telephone (baseline and 2‐monthly to discuss whatever surrogate raised; each call mean 10 minutes) | Usual care (no contact with nurse, QPL, oncologists cueing to endorse QPL or question asking) |
| Provider (and training) and Recipient | Nurse‐led
Trained as PCPC: organisation, facilitation and follow‐up of CC (family, multidisciplinary staff, external health professionals)
Recipient: family members, in conference with healthcare professionals and residential home staff | Clinicians
No training described
Recipient: patients, surrogates, clinicians | Clinicians
Training: 2.5 hour programme, small groups. Included demonstration and discussion of SICG, role play with feedback. Additional feedback after first SICG plus support (as needed)
Recipient: patients, clinicians | Physicians
No training described Recipient: patients, carers | Physicians, patient coaches Training: 3 days; included instructors in role of advanced cancer patient, role play, supporting materials
Patient training: QPL with coaching to identify most important questions/concerns/priorities
Recipient: patients, carers, physicians | Physician‐led family conference
Training: intervention meeting for investigators at each ICU site; copy of VALUE guideline. Member of study team visited each site to discuss guidelines and ensure differences between intervention and UC understood
Recipient: family members | PCT physician, social worker
Training: structured meeting elements reviewed in training session
Recipient: surrogates | Trained senior nurses
Training: 2 nurses; each receiving 40 hours’ training
Recipient: patient/carer Oncologist |
| When and how much | Single session
Timing: variable, median 48 minutes (IQR 30 to 60)
| Patients received 1‐page patient‐specific feedback via mail, to review with surrogate before consultation
Patient‐specific feedback provided to clinician and patients on day of visit (for use during consultation) | Intervention delivered over 1 or more consultations. Pre‐visit letter introducing SICG topics sent to patient (activate and prepare them for conversation)
SICG used at consultation. Clinicians triggered by researchers to have SICG conversation (by email day before/ study materials on consultation day) | Once
Participants received QPL 20 to 30 minutes before PC physician consultation | Physician training: 2 educational outreach sessions. 1st session 1 hour; 2nd booster session 45 minutes 1 month later Patients/carers: coaching session (approx. 35 to 40 minutes) 1 hour prior to consultation; follow‐up phone calls (up to 3 monthly intervals) | Once, following 3 information meetings Proactive communication conference, conducted via guidelines. Planned in advance; included senior and junior physicians, nurses, psychologist, family and friends
Family members given bereavement information leaflet, content explained orally | Once, structured face‐to‐face meeting at care facility; mean duration 47 minutes (range 20 to 75 minutes)
Social worker contacted surrogates at baseline and every 2 months via telephone | Once, face‐to‐face session with nurse, 45 minutes, 1 week before follow‐up oncology consultation; private room. Carers attended where possible
Follow‐up telephone (booster) 1 to 2 weeks after consultation following CSP delivery; 15 minutes
Oncologists: verbally cued by nurses immediately before consultation following CSP; plus postcard |
| Tailoring | Yes; topics tailored to those important to the resident
Discussion topics tailored to what was important to the resident. Could include care planning, current/future treatment decision‐making, information sharing, residents’ needs/preferences, ACP
Agenda set with input from family members and staff involved in case | Yes; patient‐specific feedback included patient‐specific highest‐ranked barrier and facilitator to EoL communication, with introductory sentence for clinician use to lower threshold to start conversations; patient’s 3 most important preferences for EoL experiences | Unclear; clinicians could split conversation across consultations but required to continue until all EMR module questions complete | Yes; QPL purpose to assist patients to identify questions of most importance and to raise these in consultation | Tailoring: oncologists’ training was individualised. Coaching tailored to patient/carer priorities and concerns for upcoming consultation | Not stated explicitly; family members had opportunity to ask questions, discuss treatment options (both intervention and UC groups) | Yes; meetings and follow‐up phone calls aimed to cover issues surrogates wished to discuss | Yes; QPL explored with patients to identify priority questions and discuss skills for question asking |
| Content | Predefined clinical triggers for CC; shared agenda setting model (resident, family, multidisciplinary team); required attendance by resident and/or family/decision makers; facilitation by PCPC to ensure optimal participation by attendees; communication strategy to summarise CC actions and plan | Patient‐specific feedback form generated from patient questionnaire responses, selected automatically (computerised process)
Selected responses included: whether their physician would know what care they would like, desire for communication about ACP, patient‐specific barriers and facilitators to communication about EoL care, preferences for CPR | SICG guide for clinicians in values and goals conversations, 7 elements: illness understanding, decision‐making and information preferences; prognostic disclosure; patient goals and fears; views on acceptable function and trade‐offs; desires for family involvement
Clinician documents discussion outcomes via structured EMR form (reminds clinicians of key discussion elements, enables documentation, able to be accessed by other clinicians). Family Communication Guide provided at consultation, suggesting approach for discussing illness/care preferences with family | QPL: 16‐page A5 booklet containing 112 questions grouped into 9 topics encompassing issues that may be discussed with physician or another health professional | Physician and patient interventions focused on same 4 elements of patient‐centred communication: engaging patients in consultations, responding to emotions, informing patients about choices for treatment and prognosis, and framing information in a balanced way | Proactive family communication conference; information on diagnosis, prognosis, treatment and discussed appropriateness of treatment limitations with family. Intensivist leading conference sought to achieve values represented by VALUE mnemonic (Value and appreciate things family says, Acknowledge emotions, Listen, ask questions that allow Understanding of who the patient is as a person, and Elicit questions from family)
Family bereavement information leaflet: 15 pages explaining EoL care, possible reactions after death of a family member, how to communicate with other family members, where to find help | Structured meeting to provide information and support to surrogates, including about treatment decisions that may arise with worsening dementia severity. PCT available for further information/assistance with decision‐making; only 3 surrogates requested additional information
Social worker follow‐up contacts provided support, presented opportunity for surrogates to raise concerns and designed to continue discussions about issues raised in face‐to‐face meeting | Face‐to‐face session based on QPL, introduced by nurse. QPL systematically explored to identify questions (including prognosis, treatment options and decisions, palliative care, patient and family support, ACP and carer‐specific issues). Prognosis and EoL care issues highlighted, skills for question asking discussed
Participants given DVD on ACP and documenting wishes for care relevant to New South Wales
Participants prompted to identify 1 to 3 questions to ask at next consultation
Follow‐up (booster) phone call; 15 minutes 1 to 2 weeks post‐consultation. Sought to reinforce face‐to‐face meeting content, prepare patients for future consultations using QPL
Nurses verbally cued oncologists prior to consultation; oncologists received postcard with suggested endorsement phrasing |
ACP: advance care plan; CC: case conference; CSP: communication support programme; EMR: electronic medical record; EoL: end of life; ICU: intensive care unit; IQR: interquartile range; PC: palliative care; PCPC: palliative care planning co‐ordinator; PCT: primary care team; QPL: question prompt list; SICG: Serious Illness Conversation Guide; UC: usual care.
*Co‐intervention(s): ICUs were part of the FAMIREA study; providing 3 formal early information meetings for all families. First meeting 24 hours (general information on diagnosis, prognosis, treatments) plus information leaflet; second meeting 48 hours (answering questions, additional information check family understanding of situation); third meeting day 3 to 5 (treatments and prognosis explained, family questions answered).
If patient was expected to die (after 3 meetings) or shift to palliative care was indicated an EoL conference was held (i.e. intervention or routine conference). Co‐interventions involved extensive information provision; authors note this may have lessened differences between intervention and UC groups for some outcomes.
Interventions included both simple and complex approaches (single or multicomponent). For instance, in Clayton 2007 patients and carers received a question prompt list (QPL) as a written booklet just prior to a palliative care physician consultation; while in the trial by Epstein 2017 a patient QPL integrated within a coaching session and physician training (focusing on the same four elements of patient‐centred communication) were tailored to each group, and both participant groups received booster/follow‐up sessions or calls to reinforce the initial session. Almost all interventions were tailored to the participants, whether by allowing patients/family/carers to nominate or guide discussions towards topics of priority, by providing patient‐specific feedback to physicians, or enabling patients/family/carers to choose or prioritise questions for discussion in consultations with physicians.
All interventions were designed to be delivered as one‐off interventions, sometimes with the addition of a booster session or follow‐up by telephone. This depended in part on the purpose of the intervention, which varied across trials. Most interventions aimed explicitly to improve patient‐doctor communication, whether by targeting both parties’ communication or knowledge or both, upskilling patients/carers to be meaningfully involved in the consultation and to ask questions or identify priorities for discussion, or by providing a face‐to‐face forum for discussions about end of life care to happen between patients, family members/carers and doctors. In some cases (e.g. Bernacki 2019) the intervention could be delivered more than once over the course of the trial, reflecting the conversation‐based nature of the intervention.
Co‐interventions were delivered only in Lautrette 2007. Here, ICUs were participants of the FAMIREA study which provided a series of early information meetings for all families of ICU patients.
Outcomes and outcome measures
Outcomes for all primary outcome categories sought by this review were reported by the included studies, as were those for all but two secondary outcome categories (health practitioner knowledge and understanding of patient/family/carer knowledge, wishes, or preferences; health systems impacts hospital admissions and re‐admissions). However, outcomes reported within categories were often disparate, timing of assessment highly variable, and outcome measures rarely compatible with one another across studies. See Additional Table 3, Table 4; Table 5; Table 6; Table 7; Table 8; and Table 9 for details of outcomes reported within each review outcome category.
| Primary outcome: knowledge and understanding
| ||||
|---|---|---|---|---|
| Study ID | Information needs | Information preferences | Shared understanding | Timing |
| X Questionnaire; total score out of 11 tallied for items not discussed, items for which they did not receive enough information, or about which they received too much information | X Questionnaire; amount of detail preferred, 1 item, Cassileth Information Styles Questionnaire 5‐point Likert scale | ‐ | 24 hours post‐consultation, 3 weeks post‐consultation | |
| ‐ | ‐ | X Discordance between prognosis ratings Researcher‐administered questionnaire/interview; 7‐point scale; ‘discordance’ defined as difference of ≥ 2 between category ratings | Shortly after audio‐recorded consultation | |
| ‐ | X Preferences for amount and type of information Self‐administered, Cassileth Information Styles Questionnaire; scores subtracted from baseline preference scores; differences expressed dichotomously (preferences met or exceeded score > 0; unmet < 0) | ‐ | 1 month | |
X: outcome assessed; ‐: outcome not reported.
The study by Lautrette 2007 reported outcomes related to knowledge (ratings of time allocated for information provision, information clarity, and information seeking by family). As these were assessed at 90 days after the death of the patient, following delivery of a one‐off family conference intervention, we judged that these outcomes were too far removed in time from the intervention to be meaningful. Data were not extracted for analysis from this study, but data are provided in Additional table 9, for transparency.
| Primary outcome: evaluation of the communication
| ||||||
|---|---|---|---|---|---|---|
| Study ID | Quality of communication | Satisfaction with consultation (communication) | Patient‐centred communication | Preferences for involvement | Goal‐consistent care | Patient‐physician relationship |
| ‐ | ‐ | X Person‐centred approach to care Care and Activities and Interpersonal Relationships and Interactions domain of Person‐Centred Environment and Care Assessment Tool; 18 items, each rated 0 (not at all) to 3 (all of the time); rated by observation, resident and family reports and documentation No data | ‐ | ‐ | ‐ | |
|
| X Quality of Communication questionnaire, (scored 0 to 100, higher better); 2 weeks | ‐ | ‐ | ‐ | ‐ | ‐
|
|
| X Quality of Communication scale, timing, scoring unclear No data | ‐ | ‐ | ‐ | X No goals met. Life Priorities Survey (patients), Family Perceptions Survey (surrogates). Baseline, 2‐monthly. Scored by matching patient final Life Priorities survey (within 3 months of death) to Family Perceptions; score 0 to 3 goals met | X Therapeutic Alliance. Human Connection Scale. Baseline, 14, 24 weeks. Total score 7 to 28; higher better |
|
| ‐ | X Questionnaire, Roter and Korsch; 25‐item scale (25 to 125); 24 hours and 3 weeks post‐consultation; higher score better | ‐ | X Actual versus preferred involvement in consultation Questionnaire 24 hours post‐consultation 5‐item rating scale (ranging from doctor leads decisions to patient leads decisions) | ‐ | ‐ |
|
| ‐ | ‐ | X Composite patient‐centred (patient‐doctor) communication Composite of 4 communication measures; coded consultation; first visit after coaching session (intervention) or study entry (control) | ‐ | X Decision regret (family) Modified decision regret scale; 8 items, 2 months post‐mortem | X Patient‐physician relationship Human Connection Scale, Health Care Communication Questionnaire, Perceived Efficacy in Patient‐Physician Interactions scale; 2 to 4 days after audio‐recorded consultation, then quarterly |
|
| ‐ | ‐ | ‐ | X Control preferences (doctor/patient +/‐ carer involvement in decisions) Self‐reported questionnaire; baseline and 1 month; Degner Control Preferences Scale; scores subtracted from baseline, differences dichotomised, preferences met/exceeded score > 0; unmet < 0 | ‐ | X Patient Communication Self‐Efficacy Self‐reported questionnaire; baseline and 1 month; Perceived Efficacy in Patient/Physician Interactions Scale |
X: outcome assessed; ‐: outcome not reported.
Satisfaction with the intervention was reported by both Clayton 2007 and Walczak 2017, but only for the intervention arm. This was therefore not extracted and reported in the review.
| Primary outcome: discussions of EoL/EoL care e.g. frequency, length, type, participants | |||||||
|---|---|---|---|---|---|---|---|
| Study ID | Discussion with clinicians | Discussion with surrogates | Documented discussion EoL care planning | Number of questions in consultation | Conversation duration | Communication content | Conversation timing |
| ‐ | ‐ | ‐ | ‐ | X Facilitated case conference intervention duration (non‐comparative) | ‐ | ‐ | |
|
| X (at last visit) Self‐reported questionnaire; 2 weeks after consultation | X (since last visit) Self‐reported questionnaire; 2 weeks after consultation | ‐
| ‐ | ‐ | ‐ | ‐ |
|
| ‐ | ‐ | X Conversation number/patient EMR review; after death | ‐ | X Physician report, post‐consultation (non‐comparative) | X Conversation content/quality SIC domains/patient EMR review; after death; coded 0 to 4 on number of domains discussed and documented (≥ 1 SIC, discussion about: values/goals, prognosis/illness understanding, EoL care, life‐sustaining treatment preferences) | X Timing of first documented SIC before death EMR review; after death |
|
| ‐ | ‐ | ‐
| X Questions, concerns, items tallied across 9 QPL categories; post‐consultation; coded 112 questions QPL grouped in 9 categories; 85 issues covered by QPL | ‐ | ‐ | ‐ |
|
| ‐ | ‐ | ‐
| ‐ | X At time of family conference with ICU staff | ‐ | ‐ |
|
| ‐ | ‐ | ‐ | X Number of questions, cues from patients, carers Coded audio‐recorded consultation; 1 week post‐CSP session; coding to identify number of direct questions, cues plus those on prognosis, EoL care, future care options, general issues | ‐ | ‐ | ‐ |
X: outcome assessed; ‐: outcome not reported.
CSP: communication support programme; EMR: electronic medical record; EoL: end of life; ICU: intensive care unit; QPL: question prompt list; SIC: serious illness conversation.
| Secondary outcomes: health practitioner knowledge and understanding; evaluation of communication, communication encounter or preparedness to communicate | |||
|---|---|---|---|
| Study ID | Evaluation of communication | Knowledge and understanding | Preparedness to communicate |
| ‐ | Staff attitudes to, knowledge of providing palliative/EoL care PCPC assessed after training; other facility staff assessed after training by PCPC; Palliative Care for Advanced Dementia tool, 35 items (qPAD) No data | Staff confidence in providing palliative/EoL care PCPC assessed after training; other facility staff assessed after training by PCPC; Palliative Care for Advanced Dementia tool, 35 items (qPAD) No data | |
| ‐ | ‐ | Uptake and effectiveness of clinician training Use of conversation tool | |
| Physician satisfaction with consultation 24 hours and 3 weeks post‐consultation | ‐ | ‐ | |
Bernacki 2019 also reported measures of clinician uptake and effectiveness of training to use their tool, and use of the tool. These were judged as measures related to implementation, rather than effects, of the intervention and data were therefore not analysed in this review.
EoL: end of life; PCPC: palliative care planning co‐ordinator.
| Secondary outcomes: health systems impacts (quality of EoL care, ratings of concordance with patient preferences for EoL care) | |||
|---|---|---|---|
| Study ID | Quality of EoL care (nurse rated) | Quality of EoL care (patient/family rated) | Ratings of concordance with patient preferences |
|
| X Nurse ratings of · CAD‐EOLD (higher scores = more comfort) · SM‐EOLD (higher scores = lower symptom frequency) Face‐to‐face/telephone interview; as soon as possible after patient death | X Family ratings of · CAD‐EOLD last 7 days life (higher scores = more comfort) · SM‐EOLD last 90 days of life (higher scores = lower symptom frequency) · SWC‐EOLD last 90 days of life (higher scores = more satisfied) Face‐to‐face/telephone interview; 4 to 6 weeks after patient death | ‐ |
|
| ‐ | ‐ | X PEACE scale, 2 subscales: · Peaceful Acceptance of Illness (acceptance of diagnosis, inner calm, feelings of being well‐loved); 5 questions, total score 5 to 20 · Struggle with Illness (feelings of upset, worry, anger, etc.), 7 questions, total score 7 to 28 Baseline, 2‐monthly |
|
| ‐
| X Caregiver evaluation of quality of EoL care 2 months post‐mortem, 6 items | ‐ |
|
| ‐ | X Surrogates' ratings of · SM‐EOLD last 90 days of life (higher scores = lower symptom frequency) · SWC‐EOLD last 90 days of life (higher scores = more satisfied) · Satisfaction with care (higher scores = better care) Interview, questionnaire; baseline, 3 and 6 months | ‐ |
X: outcome assessed; ‐: outcome not reported.
CAD‐EOLD: Comfort Assessment In Dying with Dementia; EoL: end of life; SM‐EOLD: Symptom Management at the EoL in Dementia; SWC‐EOLD: Satisfaction with Care at EoL in Dementia.
| Secondary outcomes: health systems impacts (costs of care, hospital (re)admissions) | ||||||
|---|---|---|---|---|---|---|
| Study ID | Costs | Hospital (re)admissions | Consultation length | Treatments and hospice use | Timing (longest follow‐up) | Scale, scoring |
| X Training, CC and routine healthcare costs | ‐ | ‐ | ‐ | ‐ | Cost utility (benefit estimated as QALYs). QoL for economic analyses assessed by nurse‐rated EQ‐5D‐5L No data | |
| ‐ | ‐ | X | ‐ | ‐ | ‐ | |
| ‐ | ‐ | ‐ | X | ‐ | Treatments and hospice use in last month of life; medical records, composite score of 3 indicators of aggressive treatment in last 30 days of life: chemotherapy, potentially burdensome interventions, ED/hospital admission) and hospice utilization | |
| ‐ | ‐ | X | ‐ | ‐ | ||
X: outcome assessed; ‐: outcome not reported.
Data from Epstein 2017 (composite of treatments and hospice use in last month of life) was judged as clinical, rather than primarily related to communication; data were not extracted on this outcome for the review.
CC: case conference; ED: emergency department; QoL: quality of life; QALY: quality‐adjusted life‐year.
| Secondary outcomes: patient/family member/carer quality of life | |||
|---|---|---|---|
| Study ID | Patient quality of life | Timing (longest follow‐up) | Scale, scoring |
| X | 3‐monthly | Quality of life in Late‐stage Dementia (QUALID); 11‐item scale. Nurse‐rated No data | |
| X | Unclear | SF‐12 V2 health survey (QoL and general physical health function) No data | |
| X | 3‐monthly to 3 years | Composite QoL score as average of 5 z‐scored subscales: McGillQoL scale single item, McGill Psychological Well‐Being subscale, McGill Existential Well‐Being subscale, FACT‐G Physical Functioning subscale, FACT‐G Social Functioning subscale | |
| X | 1 month | Health‐related QoL (FACT‐G) | |
X: outcome assessed.
QoL: quality of life.
For instance, the primary outcome category of knowledge and understanding was reported by four of the eight included studies (Additional Table 3). Two studies reported 'information preferences' (Clayton 2007; Walczak 2017), assessed with similar tools and at comparable time points, but other outcomes within this category were each reported by only a single study (information needs, shared understanding, time for information, information clarity, additional information sought).
Similarly while six of eight studies reported outcomes within the category 'evaluation of the communication' (Additional Table 4), there was little comparability of measures across studies. For example, four studies reported measures of patient‐centred communication: one with no data available (Agar 2017), one reporting a composite of four measures (Epstein 2017), and the others reporting measures of patients' control preferences or desire for involvement in the consultation using different scales (Clayton 2007; Walczak 2017). Even in cases where outcome measures were comparable, data were often sparse or unavailable. As an example, quality of communication was assessed by two studies (Au 2012; Bernacki 2019), both using the quality of communication questionnaire but with data available only for one study (Au 2012).
Such differences in outcome measures across studies and within outcome categories prevented pooling of data statistically in some cases. Instead, where data could not be pooled, studies were grouped according to outcomes and synthesis was conducted descriptively.
No included studies reported outcomes relating to health practitioner knowledge and understanding of patient/family/carer knowledge, wishes, or preferences, or hospital admissions and re‐admissions (Additional Table 7). Epstein 2017 reported measures of treatments and hospice use in the last months of life, assessed via medical records and through calculation of a composite score of three indicators of aggressive treatment in last 30 days of life: chemotherapy, potentially burdensome interventions, emergency department (ED)/ hospital admission) and hospice utilisation. We judged these outcomes as clinical in focus and data were therefore not extracted for analysis in this review.
For several other outcomes, data were not reported, or were not yet available in a form that was usable for this review. This included health professionals' knowledge, attitudes and confidence, quality of life, costs and person‐centred approach to care (Agar 2017), and perception of the quality of communication and quality of life (Bernacki 2019). Additionally, outcomes with data for the intervention arm only, such as satisfaction with the intervention (Clayton 2007; Walczak 2017) were not analysed for this review.
We explored groupings of studies and the possibility of undertaking meta‐analysis to pool results where possible. It was not possible to conduct meta‐analysis, and we therefore conducted a descriptive synthesis of results without statistical pooling.
Excluded studies
Studies assessed in full text but excluded from the review are reported in the Characteristics of excluded studies table, with major reasons for exclusion. Studies were excluded most commonly for the following reasons: wrong intervention or no communication intervention (62 studies), population not at EoL according to the review's definition (49 studies), focus on advanced care planning/advanced directives and uptake (35 studies), focus of the study being primarily clinical management rather than communication (19 studies), wrong study design (19 studies), focus of the study on health provider communication skills training (19 studies), or the intervention was not delivered in face to face format (11 studies).
Risk of bias in included studies
Risk of bias, by tool domain, is reported below. Studies were at generally low risk of bias, particularly for selection bias. Blinding (performance bias) was the most obvious source of bias, but overall the included studies did not have major methodological limitations. See Figure 2.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Allocation
Most (six of eight) studies used low risk methods (such as a computer‐generated random number sequence) to allocate participants to groups, with two studies providing inadequate details to judge the risk of bias (Au 2012; Reinhardt 2014). Allocation was adequately concealed in seven studies, using methods such as sealed, opaque envelopes to preserve the random number sequence until allocation, or an off‐site study statistician administering the sequence. One study (Reinhardt 2014) did not report enough details to make a clear judgement, and was rated as at unclear risk of bias.
Blinding
Blinding of participants and personnel was challenging in many of the included studies, due to the nature of the intervention. Five studies were rated as at unclear risk of performance bias, as it was not clear what effect any measures taken to blind participants or personnel, or the lack of blinding, might have been on outcomes assessed by these studies. Three studies (Lautrette 2007; Reinhardt 2014; Walczak 2017) were rated at high risk, stating that blinding was not possible and the lack of blinding may have affected delivery of the intervention to family members, and their interactions (Lautrette 2007); or that blinding was not possible and it was judged that self‐reported subjective outcomes may have been influenced by the lack of blinding (Reinhardt 2014; Walczak 2017).
Blinding of outcome assessment was judged as at low risk of bias for seven of the eight included studies. Only Walczak 2017 was rated as at high risk of bias as most outcomes were self‐reported and subjective (e.g. satisfaction, control preferences) and may have been influenced by knowledge of intervention assignment.
Incomplete outcome data
Participant numbers in trials from recruitment to follow‐up are given in Additional Table 1.
Four of eight studies were rated as at low risk of attrition bias, reporting low levels of losses (e.g. < 3% of questionnaire items missing reported by Epstein 2017) that were comparable or balanced across study groups (Agar 2017; Clayton 2007; Epstein 2017; Lautrette 2007). Three studies were rated as unclear risk of attrition bias: Au 2012 described withdrawals of 15% to 22% in control and intervention arms respectively, with mostly comparable reasons between groups. However, more people (15 versus 6) in the intervention arm refused to continue participation, and it is possible this may have introduced bias. Reinhardt 2014 reported 80% or more of participants completed both 3‐ and 6‐month follow‐ups. However, numbers for some outcomes were substantially lower, with some differences between study groups, and missing data may have influenced the results. Similarly, Bernacki 2019 reported patient participation rates that were low over the trial course, but comparable between study arms. However, there were some differences between those with analysable data and those without, and it is possible this may introduce bias.
One study (Walczak 2017) was rated at high risk of attrition bias, with 28% overall lost to follow‐up. This was possibly largely explained by declining health of participants. However, rates of loss were higher in the intervention group (34% intervention group versus 18% control) and no reasons for differential attrition were identified and may have introduced bias.
Selective reporting
Four of eight studies (Agar 2017; Au 2012; Bernacki 2019, Epstein 2017) had protocols available and all outcomes were reported as planned, or author contact confirmed that publication of data for outstanding outcomes is pending. The remaining studies typically reported all outcomes stated in the methods for the trial, but either we were not able to identify a protocol (Clayton 2007; Lautrette 2007; Reinhardt 2014), or outcomes stated in the protocol may be yet to be reported (Walczak 2017).
Other potential sources of bias
Most studies (six of eight) were at low risk of other sources of bias. This included all four cluster trials being assessed at low risk of selective recruitment of cluster participants.
Two studies were rated as at unclear risk (Au 2012; Clayton 2007): both reported some existing baseline differences between study arms, although the implications of these differences are uncertain.
Effects of interventions
See: Summary of findings 1 Summary of findings
All included studies assessed the effects of interventions to improve interpersonal communication in comparison with usual care. None assessed the comparative effects of different interventions.
Where possible we pooled data statistically, but due to variability in outcome categories and measures reported across studies, much of the synthesis was descriptive, and is presented in tables and text.
We present below a synthesis of results, organised by outcome categories sought by the review, and outcome subcategories within the data where appropriate.
Interventions to improve interpersonal communication versus usual care
Knowledge and understanding
Four studies reported outcomes related to knowledge and understanding, or appraisals of information received, associated with EoL or EoL care.
Three were conducted in patients with advanced cancer or other life‐limiting conditions, with or without carers present. All included the use of a question prompt list (QPL) but the interventions ranged from simple (QPL alone; Clayton 2007) to complex (QPL with physician and patient training, or delivered alongside a nurse‐led communication skills programme; Epstein 2017; Walczak 2017). None of these studies reported substantial differences between intervention and usual care groups, despite measuring a range of outcomes related to information needs, information preferences, and understanding. Differences in outcome measures prevented statistical pooling of data.
Epstein 2017 reported shared understanding of prognosis, assessed as mismatch (discordance) between patient and physician estimates of 2‐year survival and curability of the patient’s cancer. Ratings were assessed on a seven‐point scale, with discordance defined as a difference of two or more points between category ratings performed by patients and clinicians. 2‐year survival discordance was similar across study groups (intervention 59% versus usual care 62%), as was discordance of curability estimates (intervention 39% versus usual care 44%).
Walczak 2017 assessed preferences for the amount and type of information received in consultations, dichotomising the data into ‘met/exceeded’ and ‘unmet’ by subtracting scores at 1‐month follow‐up from baseline scores (preferences met/exceeded giving a score > 0, unmet < 0). Preferences for amount of information were similar between arms (met/exceeded intervention 56.5% versus usual care 57.1%), while preferences for type of information were met or exceeded more often in the intervention group (92.6%) than in the group receiving usual care (80%).
Clayton 2007, reporting information needs and preferences at 3 weeks post‐consultation, reported few differences between intervention and usual care groups. There were no differences in patients’ unmet information needs overall, although these were slightly higher in the intervention group (27% versus usual care 12%). Similarly, there were no differences in total scores (out of 11) for patients’ perceptions of whether topics were discussed, whether they had unmet information needs, or there was too much information provided. Only one of 11 topic items showed a difference between groups, with 4% (intervention) compared with usual care (16%) reporting they had not discussed the topic (‘what is happening with my illness’) (P = 0.05) with the palliative care team.
In both the Walczak and Clayton studies it is noteworthy that there were few differences between groups in information needs and preferences following the use of interventions involving a QPL at minimum; and that in both studies a substantial proportion of participants in both arms (up to approximately 44%, ranging from 12% to 43.5%) responded that they had unmet information needs.
A final study (Lautrette 2007) was conducted in ICUs, with outcomes assessed by telephone interview with surrogates 90 days after the patient's death. This trial assessed the effects of a single, short (mean 10 minutes) family conference provided in addition to usual care and a substantial co‐intervention (a series of early family information sessions) in the time leading up to the patient’s death. We judged that outcomes assessed at a time point so removed from an intervention delivered once would likely be confounded by many factors and so do not report data for this study. Data extracted on surrogates’ ratings of information provided prior to death is provided for transparency in Additional Table 10.
| Study ID | Outcome category | Outcomes reported | Results | Assessment method and timing |
|---|---|---|---|---|
|
| Knowledge and understanding | Time for information | Sufficient time Intervention 51/56 (91%) Usual care 45/52 (87%) | Surrogate telephone interview; 90 days after death of patient |
| Clarity of information | Information was clear Intervention 52/56 (93%) Usual care 45/52 (87%) | Surrogate telephone interview; 90 days after death of patient | ||
| Additional information requested by family | Additional information was requested Intervention 17/56 (30%) Usual care 24/52 (46%) | Surrogate telephone interview; 90 days after death of patient | ||
| Anxiety (moderate to severe symptoms; GAD‐7) | 10.4% (intervention) versus 4.2% (usual care) | 24 weeks post‐baseline (approximately 12 weeks post‐intervention) | ||
| Anxiety (STAI) | 38.7 (intervention) versus 37.5 (usual care) | 3 weeks post‐intervention | ||
| Anxiety (HADS, score > 8) | 44.6% (intervention) versus 67.3% (usual care) | 90 days after death of patient | ||
Data at longest time point are reported unless otherwise indicated.
HADS: Hospital Anxiety and Depression Scales; STAI: State‐Trait Anxiety Inventory.
Overall, interventions to improve communication may have little or no effect on measures of knowledge of illness and prognosis, or information needs and preferences, although studies were small and measures used varied across trials. We assessed the certainty of evidence as low, downgrading (‐1) due to inconsistency (different outcome measures and concepts assessed across studies) and (‐1) for indirectness (all participants were older patients with advanced cancer, and results may not apply to other populations nearing EoL).
Evaluation of the communication (positive and/or negative constructs)
Six studies reported some measure of quality, patient‐centredness or other measure of the communication occurring between patients and/or carers and health professionals related to EoL or EoL care. All were assessed in patients with advanced life‐limiting conditions, including advanced cancer and COPD. Data were grouped by similar outcome constructs and are reported below.
Quality of communication
Quality of communication was measured by two studies (Au 2012; Bernacki 2019) using the Quality of Communication (QOC) questionnaire, but data were available for only one. Au 2012 reported a slightly higher mean total score 2 weeks post‐consultation in those receiving the patient‐specific feedback intervention (mean 34.0, 95% confidence interval (CI) 28.5 to 39.4), compared with usual care (mean 25.5, 95% CI 20.4 to 30.5). However, as 100 was the maximum possible score on this questionnaire, it is noteworthy that mean scores remained low in both arms, suggesting that the intervention may have limited effectiveness or there may have been implementation issues affecting its delivery or both.
We are uncertain about the effects of interventions on the quality of communication. We assessed the certainty of evidence as very low, downgrading (‐1) due to methodological limitations (sequence generation was at unclear risk of bias), (‐1) due to imprecision (results are from a single, small study) and (‐1) for indirectness (almost all participants were older males with COPD, and results may not apply to other population groups nearing EoL).
Patient‐centred communication
The trial by Epstein 2017 reported a composite measure of patient‐centred communication, developed from assessment of four communication domains (engaging, responding, informing, framing of decisions), the component domain scale scores transformed to z scores, and these averaged to give an overall measure. Overall mean scores across domains were slightly higher in the intervention group (mean 0.2, SD 0.8) compared with the group receiving usual care (mean 0, SD 0.7), assessed post‐consultation. This trial also reported that carers’ decisional regret was slightly lower with the intervention (mean 16.0, SD 6.6) compared with usual care (mean 18.1, SD 7.1), assessed 2 months after the patient’s death. The level of patient‐centred approach to care (using the Care and Activities and Interpersonal Relationships and Interactions domain of the Person‐Centred Environment and Care Assessment Tool, PCECAT) was assessed in one further study (Agar 2017) but data were not available.
Overall, interventions to improve communication may have little or no effect on measures of patient‐centred communication. We assessed the certainty of evidence as low, downgrading (‐1) for imprecision (results are based on a single small study) and (‐1) for indirectness (all participants were older patients with advanced cancer, and results may not apply to other populations nearing EoL).
Preferences for involvement
Two studies reported on achievement of preferred level of involvement in consultations (Clayton 2007) or decisions (Walczak 2017). Both assessed the effects of interventions which included a patient QPL, with the Walczak study including additional coaching. Clayton reported no difference between groups in numbers of patients achieving their preferred level of involvement (intervention 44% versus usual care 41%), or for numbers of patients being either more or less involved in the consultation than was their preference. Walczak reported similar findings, with small differences between groups on two measures of involvement assessed as the difference between follow‐up (1 month) and baseline scores. Fewer people in the intervention group had their preferences met or exceeded for the amount of doctor/patient involvement in decisions (mean intervention group 55.6% versus usual care 69.6%), while more in the intervention arm had preferences met or exceeded for the amount of doctor/patient/carer involvement in decisions (intervention mean 87.5% versus usual care 80.8%).
Overall, interventions to improve communication may have little or no effect on participants’ preferred level of involvement. Certainty was rated as low, and was downgraded (‐1) for imprecision (results are based on a small number of participants) and (‐1) for indirectness (all participants were older patients with advanced cancer, and results may not apply to other populations nearing EoL).
Doctor‐patient relationship
Three studies measured effects on the doctor‐patient relationship, all assessing complex interventions (Bernacki 2019; Epstein 2017; Walczak 2017), using a variety of scales including the Therapeutic Alliance and PEPPI (Patient Communication Self‐Efficacy) scales. Time points also differed: 14‐week data, rather than that at the longest time point (24 weeks) was selected for analysis from Bernacki in order to be most comparable to time points for the other two trials (Epstein 8 weeks; Walczak 4 weeks); and the data for Bernacki was recalculated based on reported 95% CIs, using group sample sizes of 38 and 26 for the intervention and control groups respectively.
Pooled analysis of the three studies indicated little or no effect of the intervention on doctor‐patient relationships, compared with usual care (standardised mean difference (SMD) 0.23, 95% CI ‐0.06 to 0.51; I2 = 17%; 3 trials, 238 participants; Analysis 1.1).
Interventions to improve communication may have little or no effect on measures of the doctor‐patient relationship. Evidence was rated as low certainty, downgraded (‐1) for imprecision (results are based on a small number of participants) and (‐1) for indirectness (all participants were older patients with advanced cancer, and results may not apply to other populations nearing EoL).
Satisfaction with the consultation
Satisfaction with the consultation was reported by Clayton 2007, with no difference between the QPL intervention group (mean 110.1) and usual care (mean 110.3) at 3 weeks post‐intervention. Both arms reported high mean satisfaction levels, the maximum score on this scale being 125.
Interventions to improve communication may therefore have little or no effect on consultation satisfaction. Evidence was rated as low certainty, downgraded (‐1) for imprecision (results are based on a single, small study) and (‐1) for indirectness (all participants were older patients with advanced cancer, and results may not apply to other populations nearing EoL).
Summary of effects on the broad outcome category ‘Evaluation of communication’
Overall, the results indicate that there may be minimal or no effects of interventions to improve communication about EoL and EoL care, compared with usual care, on outcome domains encompassed by the broader category of evaluation of the communication. Even where outcomes were reportedly statistically significantly different between groups, such as for quality of communication, mean ratings were low across both intervention and control groups, suggesting that the interventions assessed may not have profound effects on such outcomes. All outcomes were rated as low‐ or very low‐certainty evidence. Further research is likely to change these results.
Discussions of EoL/EoL care
Duration and timing of EoL discussions
Three studies (Agar 2017; Bernacki 2019; Lautrette 2007) reported this outcome, only one reporting comparative data. Lautrette 2007 reported that the median duration of the family conference intervention was 30 minutes (interquartile range (IQR) 19 to 45 minutes), compared with usual care (median 20 minutes, IQR 15 to 30 minutes). Certainty of the evidence was rated as low, being downgraded (‐1) for imprecision (results are based on a single small study) and (‐1) for indirectness (all participants were older patients in ICU, results may not apply to other populations approaching the EoL). The intervention may therefore increase the duration of discussion of EoL care in an ICU setting, but the significance of this for practice is not clear.
Bernacki 2019, assessing the timing of the first documented Serious Illness Conversation (SIC) prior to death, reported that conversations happened substantially earlier among those in the intervention group (median 143 days prior to death (IQR 71 to 325), compared with the usual care group (71 days, IQR 33 to 166) (P < 0.001). We rated the certainty of the evidence as low, downgrading (‐1) for imprecision (as results are based on a single small study) and (‐1) for indirectness (all participants were older patients with advanced cancer and results may not apply to other populations approaching the EoL). The intervention may therefore lead to earlier discussions of EoL and EoL care, when compared with usual care.
Occurrence of discussions of EoL care
Four studies reported data on measures of conversations occurring about EoL/EoL care. In Au 2012 and Bernacki 2019 these took the form of self‐reported or documented discussions about treatment preferences (and related issues in Bernacki), with interventions in both studies aiming to increase the quality and occurrence of EoL discussions. In comparison, Clayton 2007 and Walczak 2017 assessed question‐asking in consultations following the intervention (which included a QPL, with or without additional coaching), as measures of patient and carer engagement in consultations.
Outcome data from Au 2012 (number of self‐reported discussions of treatment preferences with their clinicians at their last visit) and Bernacki 2019 (numbers of patients with at least one Serious Illness Conversation (SIC) documented prior to death) were pooled statistically, indicating that the intervention increased the occurrence of such discussions, compared with usual care (risk ratio (RR) 1.96, 95% CI 1.61 to 2.39; I2 = 0%; 2 trials, 537 participants; Analysis 1.2).
Consistent with reported discussions with their clinicians, Au 2012 also reported that patients’ self‐reported discussions with surrogates since the last clinic visit was slightly higher amongst those in the intervention group (53.6%) than in those receiving usual care (45.2%). The intervention’s primary aim was to increase the rates (occurrences) of discussions between patients and their clinicians, with discussions between patients and surrogates forming a secondary aim. Although in this study the intervention did increase the numbers of patients reporting discussions with both their clinicians and their surrogates, it is worth noting that these remained quite low overall, particularly for clinician discussions, and suggests that the intervention has limited effectiveness. That rates of discussions were higher between patients and carers is perhaps unsurprising, but here too only around half of patients reported these discussions occurring, with little difference between study groups.
Bernacki 2019 also reported other measures reflecting the occurrence of SIC. The number of documented SIC per patient were higher in the intervention arm, overall and when considered per domain; see Table b below.
Clayton 2007 and Walczak 2017 assessed question asking in consultations, reflecting the use of a QPL as a component (or all) of the intervention under evaluation. Meta‐analysis indicated that there may be little or no effect of the intervention, compared with usual care, on mean total numbers of patient questions (MD 1.58, 95% CI ‐1.82 to 4.98; I2 = 81%; 2 trials, 249 participants; Analysis 1.3). Numbers of carer questions could not be pooled statistically as numbers of carers per group were not available for both studies. Findings were mixed, with approximately twice as many carer questions with the intervention, compared with usual care, reported by Clayton 2007 (intervention mean 4.4 (SD 3.49) versus usual care (2.1 (SD3.49)), and no difference between groups in Walczak 2017 (both groups mean 3.5 questions). Heterogeneity was high with no obvious reason identified: participants were all older, with advanced cancer or another life‐limiting condition; interventions both included a QPL, one (Walczak 2017, also including additional patient coaching), and numbers of questions representing relatively straightforward measures. These outcomes (data) however were obtained via coding of consultations and tallying of questions, and it is possible that this may not be a highly sensitive and/or reliable method for quantifying the effects of these interventions. The approach taken, together with other factors we cannot yet identify, may contribute to the variability in results.
We are uncertain about the effects of interventions to improve discussions about EoL care. Overall, certainty was rated as very low, downgraded (‐1) for inconsistency (two of four studies indicated that the intervention had no real effect, with residual variation despite similar populations and interventions), (‐1) for methodological limitations (the largest study rated as unclear risk of bias on sequence generation), and (‐1) for indirectness (participants were older patients with cancer or COPD and may not apply to other populations nearing EoL).
Content of communication
Bernacki 2019 reported numbers of SIC domains per patient, reflecting coverage of the four domains covered in these EoL conversations. This was reported as a measure of conversation quality by the trial; in this review we considered it a measure of conversation content, as we cannot determine from coverage alone how well the domains were covered. Intervention group participants had more documented SIC domains per patient, overall, and when SIC domains were considered individually. Clayton 2007 reported little difference between groups for number of QPL topics discussed in consultations. For the latter it is worth noting that, for a total score of 85 topics that could be covered, mean scores remained low in both trial arms (approximately 21 or fewer). See Table b.
Interventions to improve communication may have little or not effect of amount of content communicated in consultations. Certainty was rated as low, downgraded (‐1) for imprecision (results are based on a small number of participants) and (‐1) for indirectness (all participants were older patients with advanced cancer, and results may not apply to other populations nearing EoL).
Table b
| Study | Outcome | Intervention | Usual care |
|---|---|---|---|
| Number of documented SIC per patient (overall)* | Mean 3.1 (95% CI 2.5 to 3.6) | Mean 2.1 (95% CI 1.4 to 2.8) | |
|
| 89% | 44% | |
|
| 91% | 48% | |
|
| 80% | 68% | |
|
| 63% | 32% | |
| Number of documented SIC domains per patient** | Mean 3.2 (95% CI 2.9 to 3.6) | Mean 1.9 (95% CI 1.6 to 2.3) | |
| Number of QPL topics discussed (out of 85 topics) | Mean 20.9 | Mean 17 |
*reported as statistically significant P = 0.02.
**reported as statistically significant P < 0.001.
Adverse outcomes or unintended effects
Adverse outcomes are difficult to define for this review, and may more appropriately be considered as unintended effects of the intervention. These might include less understanding (confusion) about EoL information, options or decisions; worsened ratings for quality of care or communication or both at the end of life; or heightened fear, distress or anxiety in patients, family members and/or carers. Included studies did not report any outcomes that might be considered harms directly associated with the interventions: only anxiety (patient or carer or both) was reported as a potential unintended consequence. However, anxiety is very difficult to interpret in this population group: anxiety levels are likely to be high and there are many potentially confounding factors in play. Tools must be tailored to assess anxiety appropriately, and other influencing factors also need to be adequately taken into account at the time of such assessments (e.g. treatment with opiates), so that effects on such outcomes are interpreted in a meaningful way.
Three studies assessed and reported anxiety of patients and/or carers (Bernacki 2019; Clayton 2007; Lautrette 2007). However, tools used were typically generic (e.g. State‐Trait Anxiety Inventory (STAI)), and anxiety measured at weeks or months post‐intervention. We judged these outcomes to be seriously confounded and too far removed in time from the intervention delivery to allow meaningful interpretation in relation to the intervention’s effects. The data are reported in Additional Table 10, for transparency, but were not analysed further for the purposes of this review.
No other unintended consequences, or worsening of desired outcomes, was reported by the included studies.
Health practitioner outcomes (knowledge; preparedness; communication evaluation)
Three studies reported some measure related to health practitioners’ knowledge and understanding, preparedness to communicate or evaluation of the communication or of the communication encounter. Only Clayton 2007 reported useable data: physician ratings of satisfaction with communication during the consultation were comparable across intervention and usual care groups (e.g. mean ratings of ‘very satisfied’ 28% intervention versus 23% usual care; ratings of ‘not satisfied’ 12% in both groups).
Agar 2017 assessed staff knowledge, attitudes and confidence in providing palliative/EoL care after they had received training (using the Palliative Care for Advanced Dementia tool, qPAD), but useable data were not available. Bernacki 2019 assessed measures of clinician uptake and effectiveness of training to use their tool, and use of the tool. We judged these as measures related to implementation, rather than effects, of the intervention, and data were therefore not analysed in this review.
Interventions to improve communication may have little or no effect on health practitioner outcomes. We rated certainty as low, downgrading (‐1) for imprecision (results are based on single small study) and (‐1) for indirectness (all participants were older patients with advanced cancer).
Patient/carer quality of life
Four studies assessed quality of life (QoL), all using different scales, and two without useable data (Agar 2017; Bernacki 2019). Epstein 2017 reported no differences between intervention and usual care arms using a composite scale derived from several tools, noting that QoL was stable until 6 to 9 months prior to death, from which point it declined. Similarly, Walczak 2017 reported little difference between QoL ratings on the FACT‐G scale at 1 month post‐intervention (intervention mean 70.9 (SD 16.3) versus usual care mean 77.8 (SD 18.8)).
Interventions to improve communication may have little or no effect on patient/carer quality of life. Certainty was rated as low, downgraded (‐1) for imprecision (results are based on studies with small sample sizes) and (‐1) for indirectness (all participants were older patients with advanced cancer).
Health systems impacts relevant to the effects of communication
Quality of EoL care
Outcomes related to carer or clinician ratings of the quality of EoL care were reported in three studies (Agar 2017; Epstein 2017; Reinhardt 2014). In all of these studies, family member or surrogate ratings of care were reported, and one (Agar 2017) also reported ratings by nurses caring for patients at the EoL.
Agar 2017 and Reinhardt 2014 assessed quality of care perceptions in patients with advanced dementia, reporting against some of the same subscales of the End of Life in Dementia (EOLD) tool. Data corresponding to two subscales could be pooled: Satisfaction with Care at EoL in Dementia (SWC‐EOLD) and Symptom Management at the EoL in Dementia (SM‐EOLD) scales. Higher scores on both scales represented an improvement, with outcomes measured at 4 to 6 weeks (Agar 2017) and at 3 months (Reinhardt 2014) after the patient’s death. Data were analysed at 3 months, rather than the longest time point (6 months) for the Reinhardt study, so that timing of assessments would be as similar as possible to that of the Agar 2017 and Epstein 2017 studies.
Pooled analysis indicates that there may be little or no effect of the intervention, compared with usual care, on family members'/carers’ ratings of symptom management at the EoL (MD ‐1.98, 95% CI ‐4.38 to 0.43; I2 = 0%; 2 studies, 212 participants; Analysis 1.4) or their satisfaction with care (MD 0.44, 95% CI ‐0.99 to 1.87; I2 = 0%; 2 studies, 212 participants; Analysis 1.5).
Agar 2017 also reported data for the Comfort Assessment in Dying with Dementia (CAD‐EOLD) scale, with little or no difference between groups, a pattern reflected in nurse ratings of the CAD and SM subscales measured soon after the patient’s death (see Table c below).
Surrogate/carer ratings of care (satisfaction) and of the quality of care were assessed by two studies (Epstein 2017; Reinhardt 2014), again with little difference between groups when measured 2 to 3 months after the patient’s death.
Communication interventions may have little or no effect on the quality of EoL care. We rated the certainty as low, downgrading (‐1) for methodological limitations (as one study (Reinhardt 2014) was rated at unclear risk of bias for sequence generation and allocation concealment) and (‐1) for indirectness (as all participants were older patients with advanced cancer).
Table c
| Outcome, scale | Study | Intervention group mean (SD) | Usual care group mean (SD) |
|---|---|---|---|
| Comfort Assessment in Dying with Dementia (CAD‐EOLD), family rated (higher scores better) | 34.7 (5.9) | 35.5 (5.9) | |
| Comfort Assessment in Dying with Dementia (CAD‐EOLD), nurse rated (higher scores better) | 32.1 (6.1) | 33.3 (5.7) | |
| Symptom Management at the EoL in Dementia (SM‐EOLD), nurse rated (higher scores better) | 22.4 (9.6) | 23.2 (8.3) | |
| Surrogates’ satisfaction with care, 3 months | 7.9 (1.4) | 7.8 (1.6) | |
| Carers’ evaluation of quality of care, 2 months | 49.6 (10) | 46.9 (9.7) |
Ratings of concordance between patient preferences for EoL care goals and care received
Concordance between goals of care and care provided at the EoL was reported by Bernacki 2019. This study also reported patients’ own ratings of their care and illness as they approached the end of life, reporting against two subscales of the Peace, Equanimity and Acceptance in the Cancer Experience (PEACE) questionnaire. This tool and its subscales assess different aspects of readiness to consider goals of care, for instance, identifying whether patients are accepting of their prognosis or whether they are still struggling to accept the approach of the end of their life. In the latter case, such struggles may need to be addressed before goals of care can be approached and discussed with the patient and their family members or carers.
This study reported little difference between intervention and usual care groups on patients’ own ratings of the two subscales: Peaceful Acceptance of Illness subscale (intervention mean 16.9 (95% CI 16.1 to 17.6) versus usual care mean 16.8 (95% CI 15.9 to 17.6)); and Struggle with Illness subscale (intervention mean 14.0 (95% CI 12.9 to 15.1) versus usual care mean 14.4 (95% CI 12.7 to 16.0)).
Concordance between goals of care and care provided at the EoL was calculated as the number of top‐3 rated goals of care identified by patients and assessed by family members as ‘met’ at the end of life (within 3 months of death). Scores could range from 0 to 3 goals met, and little or no difference between groups was reported (intervention 1.3 goals met (95% CI 1 to 1.6); usual care 1.5 (95% CI 0.9 to 2.2)).
Certainty of the evidence for both outcomes was rated as low, downgraded (‐1) for imprecision (results were based on a single small study) and (‐1) for indirectness (all participants were older adults with advanced cancer and results may not be applicable to other populations approaching the EoL).
Costs of care, hospital (re)admissions
Five studies reported health system impact outcomes but little data were presented, other than in Paladino 2020, which presented detailed healthcare utilization for the parent trial (Bernacki 2019). Agar 2017 measured costs associated with training and delivery of the intervention and usual care, and conducted a cost‐benefit utility analysis, but data were not available. Epstein 2017 reported a composite outcome assessing aspects of the aggressiveness of treatment in the last 3 months of life. However, this was judged as a primarily clinical outcome and data were not extracted for this review. Similarly, Paladino 2020b reported on measures of aggressiveness of treatment and chemotherapy receipt, and again we judged these as primarily clinical. This study also reported emergency department presentations, hospital and ICU admissions, hospice use and place of death (acute care setting), with no differences between intervention and usual care groups reported.
Both Clayton 2007 and Walczak 2017 reported consultation length, which may have implications for the costs of delivering care. In both, mean consultation length with the intervention was slightly longer (37.8 versus 30.5 minutes; 20.6 versus 20.4 respectively).
Discussion
Summary of main results
This review assessed the effects of interventions, evaluated in randomised trials, to improve or promote interpersonal communication about end of life (EoL) care between patients expected to die within 12 months, their family members and carers, and the health practitioners involved in their care. The review included all simple or complex interventions (to inform or educate, support, skill, engage, or seek participation) aiming to improve communication about EoL and EoL care. Effects were sought on a range of outcomes for health consumers, practitioners and systems, including unintended (adverse) outcomes.
Eight trials were included. All assessed the effects of interventions compared with usual care. Certainty of the evidence for all outcomes was low or very low. More specifically.
-
All outcomes were downgraded for indirectness based on the review’s purpose. Populations assessed by included trials were limited to older adults (60 years and older), conducted in urban settings in high‐income countries.
-
Outcomes were also often downgraded for imprecision or inconsistency or both. Methodological limitations were not a common reason for downgrading certainty.
A summary of the findings of the review is as follows.
Knowledge and understanding (four studies, low‐certainty evidence; one study without usable data): interventions to improve communication may have little or no effect on knowledge of illness and prognosis, or information needs and preferences, although studies were small and measures used varied across trials.
Evaluation of the communication (six studies measuring several constructs (communication quality, patient‐centredness, involvement preferences, doctor‐patient relationship, satisfaction with consultation), most low‐certainty evidence): across constructs there may be minimal or no effects of interventions to improve communication about EoL, and uncertainty about effects on quality of communication.
Discussions of EoL or EoL care (six studies measuring selected outcomes, low‐ or very low‐certainty evidence): interventions to improve communication may increase duration of EoL discussions in an intensive care unit (ICU) setting, and may lead to earlier discussions of EoL and EoL care (each assessed by one study). We are uncertain about effects on occurrence of discussions and question asking in consultations, and there may be little or no effect on content of communication in consultations.
Adverse outcomes or unintended effects (limited evidence): there is insufficient evidence to determine whether there are adverse outcomes associated with communication interventions for EoL and EoL care. Patient and/or carer anxiety was reported by three studies, but judged as confounded. No other unintended consequences, or worsening of desired outcomes, were reported.
Patient/carer quality of life (four studies, low‐certainty evidence; two studies without useable data): interventions to improve communication may have little or no effect on quality of life.
Health practitioner outcomes (three studies, low‐certainty evidence; two without usable data): interventions to improve communication may have little or no effect on health practitioner outcomes (satisfaction with communication during consultation; one study); effects on other outcomes (knowledge, preparedness to communicate) are unknown.
Health systems impacts: communication interventions may have little or no effect on carer or clinician ratings of quality of EoL care (satisfaction with care, symptom management, comfort assessment, quality of care) (three studies, low‐certainty evidence). Interventions to improve communication may have little or no effect on patients' self‐rated care and illness, or on numbers of care goals met (one study, low‐certainty evidence). Communication interventions may increase mean consultation length (two studies), but other health service impacts (e.g. hospital admissions) are unclear.
Overall completeness and applicability of evidence
The results of this review are inconclusive across the range of consumer, provider, and health system outcomes sought. This compares with the generally positive effects of interventions reported by most of the included trials. The scope of this review was, however, very broad and there are significant gaps in the evidence assembled from available trials. These gaps contributed to the generally low level of certainty we have in the results, and represent areas where future research might productively focus.
Included populations
Since EoL affects everyone, this review aimed to evaluate the effects of interventions to improve EoL communication for any and all people. Studies were therefore eligible for inclusion across the life span (neonatal to old age), in diverse settings (rural and remote, low‐ and middle‐income countries (LMIC), community, acute and chronic care), in ethnically diverse groups (e.g. culturally and ethnically diverse backgrounds), and reflecting the variability of the community (i.e. those of varying socioeconomic, health literacy, and educational status), including vulnerable or hard‐to‐reach groups (e.g. those experiencing homelessness). However, populations evaluated in trials were limited, all studying older adults in high‐income, urban settings. All outcomes were therefore downgraded for indirectness across the board, primarily because of the limited population group studied collectively, and represents a limitation – and gap for future research – in this literature.
This review defined the population of interest very broadly, but used a focused definition of EoL (a person expected to die within 12 months). Several studies were excluded on this basis (participants’ life expectancy was far longer e.g. 24 months). We adopted this criterion based on the ACSQHC 2015 consensus statement definition. However, this decision narrowed the review’s scope, and led to the exclusion of a number of studies that otherwise may have contributed data to the review. Future researchers may wish to consider this issue further if conducting studies or reviews in this area.
Related to this issue, we deliberately excluded advance care planning (ACP) from this review, as it may not be closely temporally linked to EoL. For example, older adults are increasingly encouraged to undertake ACP in preparation for future decisions. In some cases, this may be close to the end of the person’s life. In others, ACP is for an unspecified future occurrence that may be many years into the future. Due to this variability and the lack of immediacy of the decision‐making associated with ACP – as well as the heavy reliance on checklists and similar tools, and the strong focus on ACP uptake (rather than communication outcomes) – we excluded these studies. In comparison, voluntary assisted dying (VAD), was eligible for inclusion, although no relevant studies were identified. VAD focuses on decisions to end life within a short (6‐ to 12‐month) time frame, and so was consistent with the broader focus of this review. It is possible, for those trials excluded based on their ACP focus, that there may have been a small number of studies that included conversations about EoL within the final 12 months of life that were excluded from this review. We did not identify such situations during study screening for the review, but it is possible that these were not identified and this issue should be considered further in future updates to this review, or future related reviews.
Interventions evaluated
All communication purposes were eligible for inclusion. Most had a stated aim of improving some aspect of patient‐doctor communication, targeting patients and carers alone or together with practitioners. Almost all interventions were tailored to participants, whether by allowing patients/carers to nominate or guide discussions towards priority topics, by providing patient‐specific feedback to physicians, or enabling patients/carers to prioritise questions.
More specifically, interventions most often aimed to provide information to participants, or to support engagement in consultations or decision‐making. Less often were interventions focused on checking people’s knowledge and understanding; on eliciting preferences, views and goals of treatment and EoL care; or on determining people’s concerns about EoL care and communication, including those of the health practitioners involved in these communications. All of these represent areas worthy of further exploration.
Interventions included in the review ranged from the simple (e.g. question prompt list (QPL)) to the complex (e.g. QPL plus patient coaching plus health practitioner training). Almost all were delivered once, sometimes with an additional booster or brief follow‐up. Evaluating the effects of communication delivered over time is undoubtedly challenging. However, people’s priorities, preferences and understanding change over time as death nears, and communication that is responsive and tailored to people’s changing needs suggests that future studies may also fruitfully explore the delivery of communication interventions over time for people in the EoL period.
The review did not find any head‐to‐head comparisons between interventions. Future studies might consider such comparisons, along with stepwise addition of components to communication interventions, in order to systematically assess the relative effects and contributions to outcomes that might follow use of such strategies. Usual care was in all cases the control arm for the included trials, but the exact nature of this varied considerably. In some cases, usual care was standard clinical care, in others there was substantial information or support or both provided (or even a substantial co‐intervention delivered to both arms). This may have effectively narrowed any differences between intervention and usual care arms in at least some of the included trials. Future studies might consider this issue carefully and design studies accordingly.
Outcomes and outcome measures
This review sought information about the effects of interventions on a wide range of outcomes. Collectively, included studies reported outcomes across most broad categories of interest. This indicates researchers’ understanding of the complex effects of communication with different purposes related to EoL, and that different people are involved in such communication. Despite this, outcome measures (tools or timing or both) within categories were often highly varied. In general, few studies contributed data to any one outcome category or construct, so many were sparsely populated with data. As a result, findings were largely inconclusive. There were several outcomes where there were little data on which to base conclusions (e.g. those for which only a single small study contributed data); in other cases there were no data available or in usable form or both (e.g. costs). There were also a small number of specific gaps (health practitioner knowledge and understanding of patient/family/carer knowledge, wishes, or preferences; hospital admissions and re‐admissions). Despite our extensive searches for relevant evidence, it is possible that studies focused on cost‐effectiveness or health service use related to communication at the EoL were not identified for consideration for the review, and this may represent a limitation of the review.
Variability in outcome constructs and measures used in this research literature limits conclusions that can be made across studies, or pooling of data to identify whether effects of interventions exist, or both. Future studies might consider using ‐ or developing – validated, responsive tools to assess outcomes, to facilitate analysis and interpretation of findings by clinicians and by researchers.
Challenges of measuring the different constructs affected by EoL communication may also have contributed to the limited effects of the interventions across outcomes (Brighton 2016; Sansoni 2014). Such communication is complex (multidirectional, multifaceted, involving multiple people) and without a range of well‐established tools available, trialists may be left to rely on those used historically or to develop their own tools. Additionally, some outcome measures may not have been sensitive and/or specific enough to capture nuanced differences between intervention and control groups (e.g. coded qualitative data from audiotaped consultations converted to quantitative count data). In some cases, composite scores were reported, which can be difficult to interpret. In others, measures may have been a little blunt e.g. occurrence of discussions and coverage of topics may not reflect quality of the communication (how well was it done?), rather it more closely reflects content (were these topics covered?). Such challenges of measuring outcomes for EoL communication have been noted by other researchers (Brighton 2016; Sansoni 2014). Current results suggest that more sophisticated and nuanced ways of assessing communication at EoL may help to better understand the complex interactions between the people involved. It may be challenging to design studies to measure such outcomes, but continuing to conduct trials without appropriate and sensitive outcome measures will not optimally fill gaps in our knowledge or readily inform policy and practice decisions in this area.
Consideration should be given to the value of mixed‐methods or qualitative research or both, which may be better placed to inform decisions about outcomes that are meaningful to those involved in EoL consultations and discussions. Rather than rushing to replicate trials, research that explores a range of perspectives about communication at EoL may help to better understand the context in which communication is occurring, as well as barriers and enablers that may influence the complex interactions between the people involved in communication at the EoL.
Unintended (adverse) outcomes
Adverse outcomes are difficult to define amongst this population, and therefore difficult to measure and report. There may be many confounding factors present in this context of clinical care and communication. None of the included studies reported outcomes we judged as adverse (unintended effects) of the intervention, such as worsened understanding (confusion) about EoL, EoL care or options and decisions to be made; poorer ratings of quality of care or communication during the EoL period; or heightened fear, distress, anxiety or stress in patients, family members and carers, or in health practitioners involved in care.
Only patient and/or carer anxiety was reported as a potential unintended consequence of the intervention (3/8 studies), but effects were judged as confounded and there was no indication of other possible negative effects of communication interventions. This is a gap in the evidence, reflected by other studies and reviews (Brighton 2016; Sansoni 2014) and future studies should consider carefully the range, types and measurement of potential negative consequences.
Quality of the evidence
Overall, certainty of evidence was rated as low or very low for all outcomes. We rated down (‐1) for indirectness across all outcomes, as the populations assessed by the included trials were limited, as discussed above. We commonly also downgraded the evidence based on imprecision: several outcomes were measured only by single trials, and sample sizes were typically small. In a small number of cases, certainty was downgraded based on inconsistency (knowledge, discussion occurrence), as a result of variation in constructs and measures or persistent differences in findings despite similar population and intervention.
Methodological limitations (risk of bias) were not a major reason for downgrading across outcomes. Most trials were of good methodological quality, particularly on the key domains of sequence generation and allocation concealment, despite the challenges of conducting research in this area.
Potential biases in the review process
We used standard Cochrane methods to undertake this review, with few changes from protocol to review stage. The small changes made when conducting the review (changes to grey literature sources, decision to extract data at longest follow‐up) are unlikely to bias the results of the review.
Similarly, we searched extensively for relevant published and unpublished research, and conducted a range of supplementary search activities, including contacting authors of relevant trials. Despite this, it is possible that we have missed a relevant trial, or relevant publications arising from the included trials.
The review was established with detailed selection criteria that articulated several complicated distinctions in the topic area. However, making some of these distinctions operational was challenging. This included the intention to exclude studies in which ACP administration and completion were the main focus; clearly distinguishing between clinical care and communication across different clinical populations; and consistently identifying key features of the EoL patient population. These decisions were made through discussion to reach consensus amongst at least three review authors, and to ensure that decision rules were consistently applied. These screening decisions related to a large number of excluded studies (e.g. focus on ACP and uptake (30 studies excluded on this basis), population not at EoL according to the review's definition (30 studies), study focus primarily clinical management not communication (18 studies)). Others making these same selection decisions may reach different conclusions about the inclusion or exclusion of some of these studies. As a team we made every effort to apply decision rules consistently, and so do not believe that these decisions introduced bias.
Agreements and disagreements with other studies or reviews
This review’s findings ‐ which highlight uncertainty within current evidence ‐ differ substantially from those of most of the included trials, many of which individually concluded that communication interventions were effective, as assessed via a range of outcomes. Recent studies have, however, highlighted the challenges of delivering and evaluating such interventions in populations approaching EoL, and have called for rigorous evaluations of the effects of communication interventions for adult and paediatric patients and families (Ekberg 2019; Hjelmfors 2020; Wolfe 2020).
While many of the included studies reported statistically significant effects in favour of the intervention on key outcomes, the meaning of such results is in at least some cases unclear. For example, if quality of communication scores increase significantly with the intervention, but this effect is small and scores remain very much lower than the highest possible, it is very difficult to understand what this might mean for practice. Similarly, if knowledge increases significantly with an intervention relative to usual care, but 50% of people still have unmet information needs irrespective of their study group, this suggests the intervention’s effectiveness is limited.
We are not suggesting that an intervention could or should be 100% effective to be worthwhile and meaningful. However, we would urge trialists and systematic reviewers in this area to carefully consider the meaning of the findings arising from trials, particularly as they are undertaken in a vulnerable population group at a particularly stressful and distressing times of their lives. We also emphasise the importance of building carefully upon previous research to maximise the value and effectiveness of interventions for improving communication with people in the EoL period.
Such situations highlight the need for well‐established tools to measure communication interventions’ effects, and for researchers to consider the value of alternative methods for studying communication at the end of life – such as qualitative methods to unpack complexity and to understand barriers and enablers to good quality communication, or mixed methods approaches which might be harnessed to jointly understand complexity and to evaluate effects of interventions. Involvement of stakeholders in the co‐design and evaluation of interventions for communication at EoL also seems essential, and several of the included studies evaluated interventions which had been developed in just such a way. The importance of co‐design, coupled with meaningful stakeholder input to the development of interventions (and outcome measures) cannot be overstated – but the understanding of what this might involve in real terms has changed substantially over time (particularly the last decade) and is continuing to develop (Merner 2019; Merner 2021).
There are few recent systematic reviews published which have specifically focused on communication at the end of life. One new review was identified (Thode 2020), evaluating tools to support discussion of life‐prolonging treatments in hospital. This review identified a small number of studies of various designs, a summary of which suggested there may be mixed effects, including some positive effects on outcomes such as self‐efficacy. Other recent publications (Fujimori 2020; Van der Steen 2021) highlight that communication at EoL is a developing area and indicate a need for tailored approaches to communication. There is also a need to consider the complex interplay between the different people and their roles who may be involved in these discussions (Wolfe 2020), including shared decision‐making about goals for future care as diseases progress towards the end of life (e.g. for patients with dementia) (Van der Steen 2021).
Timing and timeliness of communication for EoL are factors with emerging importance in the literature, but only explored in this review to a small degree with the results of Bernacki 2019 indicating that the Serious Illness Conversation Guide (SICG) communication intervention led to much earlier discussions of EoL care. Early conversations about EoL may be better than later (Brighton 2016), although there is growing understanding that good EoL communication must take into account people’s different needs, preferences and priorities, and provide clear opportunities for preferences and concerns to be discussed, at different times (Hjelmfors 2020; NICE 2019; Thode 2020). Not all people will be ready to take part in such discussions at the same time point (Brighton 2016) and so the timing for offering information, and its staging, are critical (Anderson 2019). The length (duration) of consultations and meaning of these is also complex and not yet well understood. Sometimes longer consultations may indicate a more comprehensive discussion about EoL and care has taken place. At other times, longer meetings or consultations might indicate some level of disagreement or misunderstanding between those involved (Thompson 2009).
There must also be clear, ongoing opportunities for patients and carers to revisit and change decisions over time, tailored according to need (Anderson 2019; Brighton 2016; Ekberg 2019; Hjelmfors 2020; NICE 2019). For instance, patients and carers may have preferences for different types and amounts of information that depend at least in part on how close to the patient is to the EoL (Anderson 2019; Brighton 2016; Ekberg 2019). Cultural background and health literacy levels may also influence such preferences for information types and amounts (Anderson 2019; Hjelmfors 2020; Thode 2020). Opportunities for discussions about EoL therefore need to be provided, to account for people's varying preferences for information types and amounts (NICE 2019; Sansoni 2014). Considering such factors, likely among others, will help to determine how communication at the EoL might best take place, and whether or not it is able to meet the needs and priorities of all people involved in such discussions.
Study flow diagram.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Comparison 1: Intervention versus usual care, Outcome 1: Doctor‐patient relationship
Comparison 1: Intervention versus usual care, Outcome 2: Discussion of EoL care planning
Comparison 1: Intervention versus usual care, Outcome 3: Patient questions in consultation
Comparison 1: Intervention versus usual care, Outcome 4: Family‐rated symptom management (SM‐EOLD)
Comparison 1: Intervention versus usual care, Outcome 5: Family‐rated satisfaction with care at EoL (SWC‐EOLD)
| Communication intervention compared with usual care for end of life care | |||
|---|---|---|---|
| Patient or population: people approaching the end of life (within 12 months), their family members and/or carers Settings: any (residential care, hospital (inpatient and outpatient units) and community‐based clinics, palliative care services) Intervention: interventions to improve communication about EoL and/or EoL care Comparison: usual care | |||
| Outcomes | Intervention effects | Number of participants | Certainty of the evidence |
| Patient, family and/or carer outcomes | |||
| Knowledge and understanding
Variable scales: information (amount and type) needs/preferences; discordance between patient and physician survival and curability estimates
Timing: immediately to 1 month post‐consultation | Overall, interventions to improve communication may have little or no effect on measures of knowledge of illness and prognosis, or information needs and preferences
In 1 study (303 participants), discordant estimates of 2‐year survival between patients and doctors (intervention 59% versus usual care 62%) and curability (intervention 39% versus usual care 44%) were similar between groups (Epstein 2017). Another (79 participants) reported similar proportions of patients had their preferences for amount of information met or exceeded across intervention and usual care groups, but that type of information was met or exceeded more often in the intervention group (93% versus 80%) (Walczak 2017). 1 final study (170 participants) reported no differences in patients’ unmet information needs overall (Clayton 2007) | 552 (3 studies)c | ⊕⊕⊝⊝
|
| Evaluation of the communication: different constructs (perceptions of communication quality; patient‐centredness of communication; involvement preferences; doctor‐patient relationship measures)
Timing: immediately post‐consultation to 18 weeks post‐consultation | Across constructs (patient‐centredness, involvement preferences, doctor‐patient relationship, satisfaction with consultation), there may be minimal or no effects of interventions to improve communication about EoL and EoL care (Agar 2017; Bernacki 2019; Clayton 2007; Epstein 2017; Walczak 2017), and uncertaintye about effects on quality of communication (Au 2012)
| 6 studiesf | ⊕⊕⊝⊝ |
| Discussions of EoL/EoL care: discussion timing and length
EMR review post‐death
Timing: at time of family conference (intervention) in ICU; post‐death
| The intervention may lead to longer and earlier discussions of EoL and EoL care, compared with usual care, but each result is based on a single study
1 study (108 participants) reported comparative data: median family conference intervention duration was 30 minutes (IQR 19 to 45 minutes) versus usual care (median 20 minutes, IQR 15 to 30 minutes) (Lautrette 2007)
1 study (376 participants) reported that the first documented Serious Illness Conversation happened earlier among intervention group participants (median 143 days prior to death (IQR 71 to 325) than usual care (71 days, IQR 33 to 166) (Bernacki 2019) | 484 (2 studies) | ⊕⊕⊝⊝ |
| Discussions of EoL/EoL care: discussion occurrence
EMR review post‐death; coding of consultations; self‐reported occurrence
Timing: immediately, 1 or 2 weeks post‐consultation; after death | Overall, we are uncertain about the effects of interventions to improve discussions about EoL care
2 studies indicated that the intervention increased the occurrence of EoL discussions, compared with usual care (RR 1.96, 95% CI 1.61 to 2.39; 2 trials, 537 participants); the others indicated little or no effect of the intervention on mean total numbers of patient questions in consultations (MD 1.58, 95% CI ‐1.82 to 4.98; 2 trials, 249 participants) | 786 (4 studies) | ⊕⊝⊝⊝
|
| Adverse (unintended) outcomes
| There is insufficient evidence to determine whether adverse (unintended) outcomes are associated with communication interventions. Patient and/or carer anxiety was reported (3 studies), but was judged as confounded, and no other unintended consequences, or worsening of desired outcomes, were reported | ‐ | ‐ |
| CI: confidence interval; EMR: electronic medical record; EoL: end of life; ICU: intensive care unit; IQR: interquartile range; MD: mean difference; RR: risk ratio | |||
| GRADE Working Group grades of evidence | |||
| aDowngraded (‐1) due to inconsistency (different outcome measures and concepts assessed across studies). | |||
| Participant numbers | ||||||||
|---|---|---|---|---|---|---|---|---|
| Eligible for inclusion | Nursing homes: 111 eligible Participants: 148 UC 171 intervention | 1292 1173 mailed introductory letter | Clinicians: 133 recruited Patients: 9182 screened UC (8530 ineligible) 9395 screened intervention (8842 ineligible) | 196 | Physicians: 38 eligible Patients: 453 eligible | 132 | 214 | 363 |
| Excluded | Nursing homes: 91 (55 did not meet criteria; 36 declined) 14 UC 11 intervention | 30 screened out (did not meet criteria) 21 declined participation prior to screening | Clinicians: 6 pilot clinicians ineligible Patients: 8987 UC 9211 intervention | 22 | Patients: 137 excluded (38 ineligible, 99 refused) | ‐ | ‐ | ‐ |
| Refused to take part | ‐ | 645 (did not wish to participate) 101 (did not keep appointment) | Clinicians: 36 Patients: unclear | 18 | 99 | 6 | 104 | 253 |
| Randomised to intervention group(s) | 10 homes, 160 participants 156 received allocated intervention | 194 | Clinicians: 48 (20 clusters) Patients: 184 (20 clusters) | 92 | Physicians: 19 Patients: 139 Carers: 105 | 63 | 58 | 61 |
| Randomised to control (usual care) group | 10 homes, 134 participants 130 received allocation intervention | 182 | Clinicians: 43 (21 clusters) Patients: 195 (19 clusters) | 82 | Physicians: 19 Patients: 142 Carers: 99 | 63 | 52 | 49 |
| Excluded post‐randomisation (for each group; with reasons if relevant) | UC 66 Intervention 89 All excluded because did not die during the study period | ‐ | 96 (intervention) Death; lost to follow‐up; declined further surveys; completed 2 years 118 (UC) Death; lost to follow‐up; declined further surveys; completed 2 years Intervention: 184 (20 clusters) enrolled: 134 analysed (18 clusters); (50 total: no family/friend response, no baseline survey, withdrew) 34 (13 clusters) analysed: 74 died, 11 lost to follow‐up; 17 declined further surveys; 32 completed 2 years UC: 195 (19 clusters) enrolled: 144 analysed (17 clusters); (51 total: no family/friend response, no baseline survey, withdrew) 26 patients (13 clusters) analysed: 77 died. 12 lost to follow‐up, 21 declined further surveys; 34 completed 2 years | ‐ | ‐ | ‐ | ‐ | ‐ |
| Withdrawn (for each group; with reasons if relevant) | UC 4 Intervention 4 (died before intervention) | Control 27 (14.5%) 3 patient‐clinician relationship changed 6 refuse to continue 9 not contactable 8 no target visit 1 too ill/deceased Intervention 43 (22.2%) 4 patient‐clinician relationship changed 15 refuse to continue 8 not contactable 10 no target visit 3 too ill/deceased 3 other | Intervention 12 Control 8 | None | Intervention 6 Control 5 | Intervention 11 Control 7 | ‐ | ‐ |
| Lost to follow‐up (for each group; with reasons) | ‐ | ‐ | Intervention: 11, no reason given Control 12, no reason given | Intervention: 2, 1 mistakenly seen by a junior physician who was not participating in the study, 1 due to mechanical failure of tape recorder Control: 2, 1 mistakenly seen by a junior physician who was not participating in the study 1 due to mechanical failure of tape recorder | Intervention: 3 died Control: 1 died 1 lost to follow‐up | Intervention: did not answer telephone n = 4; experiencing severe emotional distress n = 5; refused interview n = 2 Control: did not answer telephone n = 3; experiencing severe emotional distress n = 1; refused interview n = 2; patient still alive n = 1 | Total numbers reported at each time point N = 96/110 completed 3‐month measures N = 90/110 completed 6‐month measures Plus an additional 3 where complete data were not available NB also for several outcomes data for slightly lower numbers in total are available i.e. table 3 – ranges from n = 65 to n = 81. Not clear what happened to these missing measures | Intervention: 21 (34%) Unclear; data collection hampered by declining health, attrition high (higher in intervention group) but no systematic reasons for differential dropout identified. Reasons for loss not reported specifically Control: 9 (18%) Reasons for loss not reported specifically |
| Included in the analysis (for each group, for each outcome) | Intervention 67 Control 64 These numbers were analysed throughout, with losses for particular scales/assessments noted where applicable | Intervention 194 Control 182 | Intervention 38 patients (13 clusters) Control 26 patients (13 clusters) | Intervention 90 Control 80 | Intervention 19 physicians, 130 patients Control 19 physicians, 135 patients | Intervention 52 (83%) Control 56 (89%) | 6‐month data (longest time point) Intervention 47 Control 40 | Intervention 39 Control 40 |
| Assessment of attrition bias for RoB ratings | Assessed as low risk Large proportion of data (participants randomised) missing. However, these were comparable for the 2 study groups, and was due to participants not dying during the study period (outcomes assessed for this study were focused on those around death) Withdrawal rates for other reasons were low and comparable across groups | Assessed as unclear risk Withdrawal 15% to 22% respectively control and intervention arms Reasons for withdrawal/dropout were reasonably comparable except that more (15 versus 6) refused to continue participation in the intervention group Authors report no differences in baseline characteristics regarding whether patients completed the study or were lost to follow‐up ITT analysis was used; effect of imputed data on results was examined in analysis models with authors reporting similar results where imputed and non‐imputed data were used in analysis | Assessed as unclear risk Patient participation rates and numbers analysable were low but comparable between arms | Assessed as low risk Low levels of loss to follow‐up 4/174; balanced across groups (n = 2 each), with comparable reasons | Assessed as low risk "Fewer than 3% of follow‐up questionnaires were missing" (page 95) Data seem otherwise complete for outcomes reported in main paper and in supplement 3 | Assessed as low risk Loss to follow‐up and withdrawals were acceptably low and comparable across study groups: 52/63 (83%) completed interviews at 90 days intervention group, 56/63 (89%) in control group Reasons for withdrawal/loss were similar across groups (not answering telephone, refused interview), although higher rates of severe emotional distress in intervention group (n = 5) than control group (n = 1) | Assessed as unclear risk Missing data were reported However, numbers were lower for some outcomes (such as ratings of care management) where n = 65 (numbers fairly comparable between the 2 groups) | Assessed as high risk Attrition was high (31/110 (28%) lost to follow‐up), possibly largely explained by declining health of participants (patients) Authors state that ITT analysis was used (according to group assignment) but dropout rates may be problematic |
| Additional notes | The study was underpowered for the primary outcome because fewer participants died during the study period than predicted Predicted sample size required recruitment of 272 participants (assuming 10% dropout rate) (17 per site) | ‐ | Sample size calculated as 200 evaluable patients per arm for required power, assuming 6% dropout | ‐ | ‐ | ‐ | ‐ | Trial is underpowered (sample size calculated at 140; 110 recruited; 79 completed) to detect differences between groups |
| ITT: intention to treat; RoB: risk of bias; UC: usual care. | ||||||||
| STUDY ID |
| |||||||
|---|---|---|---|---|---|---|---|---|
| Intervention(s) aim and components | To improve EoL care
Facilitated CC + patient‐centred palliative care training | To improve communication about preferences for EoL care
1‐page patient‐specific feedback form based on patient’s self‐reported responses | To evaluate the SICG
SICG + training, including support for response documentation and patient/family materials | To evaluate QPL effects on patient/carers’ EoL care topic discussions in consultations
QPL for patients and caregivers | To improve patient‐centred communication between physicians and patients/carers
Complex patient‐centred communication training (2 components: physician, patient) | To improve communication between family and ICU staff and support family decision‐making
Proactive communication family conference plus bereavement information leaflet | To provide information and support to surrogates of patients with advanced dementia
Face‐to‐face structured conversation, telephone follow‐up | To increase patients’ EoL care discussions and cue oncologists to endorse QPL and question asking
Nurse‐led CSP (QPL, booster and verbal/written cueing of oncologists pre‐consultation) |
| Comparator (usual care) | Usual care (no additional education, training, support) | Usual care (no patient‐specific feedback) | Usual care | Usual care (routine consultation with PC physician) | Usual care (oncologist met with research assistant but received no training) | Usual care (routine family EoL conference); informs family of treatment limitations and that death is imminent; led by senior physician, with at least 1 family member | Usual care plus social contact by telephone (baseline and 2‐monthly to discuss whatever surrogate raised; each call mean 10 minutes) | Usual care (no contact with nurse, QPL, oncologists cueing to endorse QPL or question asking) |
| Provider (and training) and Recipient | Nurse‐led
Trained as PCPC: organisation, facilitation and follow‐up of CC (family, multidisciplinary staff, external health professionals)
Recipient: family members, in conference with healthcare professionals and residential home staff | Clinicians
No training described
Recipient: patients, surrogates, clinicians | Clinicians
Training: 2.5 hour programme, small groups. Included demonstration and discussion of SICG, role play with feedback. Additional feedback after first SICG plus support (as needed)
Recipient: patients, clinicians | Physicians
No training described Recipient: patients, carers | Physicians, patient coaches Training: 3 days; included instructors in role of advanced cancer patient, role play, supporting materials
Patient training: QPL with coaching to identify most important questions/concerns/priorities
Recipient: patients, carers, physicians | Physician‐led family conference
Training: intervention meeting for investigators at each ICU site; copy of VALUE guideline. Member of study team visited each site to discuss guidelines and ensure differences between intervention and UC understood
Recipient: family members | PCT physician, social worker
Training: structured meeting elements reviewed in training session
Recipient: surrogates | Trained senior nurses
Training: 2 nurses; each receiving 40 hours’ training
Recipient: patient/carer Oncologist |
| When and how much | Single session
Timing: variable, median 48 minutes (IQR 30 to 60)
| Patients received 1‐page patient‐specific feedback via mail, to review with surrogate before consultation
Patient‐specific feedback provided to clinician and patients on day of visit (for use during consultation) | Intervention delivered over 1 or more consultations. Pre‐visit letter introducing SICG topics sent to patient (activate and prepare them for conversation)
SICG used at consultation. Clinicians triggered by researchers to have SICG conversation (by email day before/ study materials on consultation day) | Once
Participants received QPL 20 to 30 minutes before PC physician consultation | Physician training: 2 educational outreach sessions. 1st session 1 hour; 2nd booster session 45 minutes 1 month later Patients/carers: coaching session (approx. 35 to 40 minutes) 1 hour prior to consultation; follow‐up phone calls (up to 3 monthly intervals) | Once, following 3 information meetings Proactive communication conference, conducted via guidelines. Planned in advance; included senior and junior physicians, nurses, psychologist, family and friends
Family members given bereavement information leaflet, content explained orally | Once, structured face‐to‐face meeting at care facility; mean duration 47 minutes (range 20 to 75 minutes)
Social worker contacted surrogates at baseline and every 2 months via telephone | Once, face‐to‐face session with nurse, 45 minutes, 1 week before follow‐up oncology consultation; private room. Carers attended where possible
Follow‐up telephone (booster) 1 to 2 weeks after consultation following CSP delivery; 15 minutes
Oncologists: verbally cued by nurses immediately before consultation following CSP; plus postcard |
| Tailoring | Yes; topics tailored to those important to the resident
Discussion topics tailored to what was important to the resident. Could include care planning, current/future treatment decision‐making, information sharing, residents’ needs/preferences, ACP
Agenda set with input from family members and staff involved in case | Yes; patient‐specific feedback included patient‐specific highest‐ranked barrier and facilitator to EoL communication, with introductory sentence for clinician use to lower threshold to start conversations; patient’s 3 most important preferences for EoL experiences | Unclear; clinicians could split conversation across consultations but required to continue until all EMR module questions complete | Yes; QPL purpose to assist patients to identify questions of most importance and to raise these in consultation | Tailoring: oncologists’ training was individualised. Coaching tailored to patient/carer priorities and concerns for upcoming consultation | Not stated explicitly; family members had opportunity to ask questions, discuss treatment options (both intervention and UC groups) | Yes; meetings and follow‐up phone calls aimed to cover issues surrogates wished to discuss | Yes; QPL explored with patients to identify priority questions and discuss skills for question asking |
| Content | Predefined clinical triggers for CC; shared agenda setting model (resident, family, multidisciplinary team); required attendance by resident and/or family/decision makers; facilitation by PCPC to ensure optimal participation by attendees; communication strategy to summarise CC actions and plan | Patient‐specific feedback form generated from patient questionnaire responses, selected automatically (computerised process)
Selected responses included: whether their physician would know what care they would like, desire for communication about ACP, patient‐specific barriers and facilitators to communication about EoL care, preferences for CPR | SICG guide for clinicians in values and goals conversations, 7 elements: illness understanding, decision‐making and information preferences; prognostic disclosure; patient goals and fears; views on acceptable function and trade‐offs; desires for family involvement
Clinician documents discussion outcomes via structured EMR form (reminds clinicians of key discussion elements, enables documentation, able to be accessed by other clinicians). Family Communication Guide provided at consultation, suggesting approach for discussing illness/care preferences with family | QPL: 16‐page A5 booklet containing 112 questions grouped into 9 topics encompassing issues that may be discussed with physician or another health professional | Physician and patient interventions focused on same 4 elements of patient‐centred communication: engaging patients in consultations, responding to emotions, informing patients about choices for treatment and prognosis, and framing information in a balanced way | Proactive family communication conference; information on diagnosis, prognosis, treatment and discussed appropriateness of treatment limitations with family. Intensivist leading conference sought to achieve values represented by VALUE mnemonic (Value and appreciate things family says, Acknowledge emotions, Listen, ask questions that allow Understanding of who the patient is as a person, and Elicit questions from family)
Family bereavement information leaflet: 15 pages explaining EoL care, possible reactions after death of a family member, how to communicate with other family members, where to find help | Structured meeting to provide information and support to surrogates, including about treatment decisions that may arise with worsening dementia severity. PCT available for further information/assistance with decision‐making; only 3 surrogates requested additional information
Social worker follow‐up contacts provided support, presented opportunity for surrogates to raise concerns and designed to continue discussions about issues raised in face‐to‐face meeting | Face‐to‐face session based on QPL, introduced by nurse. QPL systematically explored to identify questions (including prognosis, treatment options and decisions, palliative care, patient and family support, ACP and carer‐specific issues). Prognosis and EoL care issues highlighted, skills for question asking discussed
Participants given DVD on ACP and documenting wishes for care relevant to New South Wales
Participants prompted to identify 1 to 3 questions to ask at next consultation
Follow‐up (booster) phone call; 15 minutes 1 to 2 weeks post‐consultation. Sought to reinforce face‐to‐face meeting content, prepare patients for future consultations using QPL
Nurses verbally cued oncologists prior to consultation; oncologists received postcard with suggested endorsement phrasing |
| ACP: advance care plan; CC: case conference; CSP: communication support programme; EMR: electronic medical record; EoL: end of life; ICU: intensive care unit; IQR: interquartile range; PC: palliative care; PCPC: palliative care planning co‐ordinator; PCT: primary care team; QPL: question prompt list; SICG: Serious Illness Conversation Guide; UC: usual care. *Co‐intervention(s): ICUs were part of the FAMIREA study; providing 3 formal early information meetings for all families. First meeting 24 hours (general information on diagnosis, prognosis, treatments) plus information leaflet; second meeting 48 hours (answering questions, additional information check family understanding of situation); third meeting day 3 to 5 (treatments and prognosis explained, family questions answered). If patient was expected to die (after 3 meetings) or shift to palliative care was indicated an EoL conference was held (i.e. intervention or routine conference). Co‐interventions involved extensive information provision; authors note this may have lessened differences between intervention and UC groups for some outcomes. | ||||||||
| Primary outcome: knowledge and understanding
| ||||
|---|---|---|---|---|
| Study ID | Information needs | Information preferences | Shared understanding | Timing |
| X Questionnaire; total score out of 11 tallied for items not discussed, items for which they did not receive enough information, or about which they received too much information | X Questionnaire; amount of detail preferred, 1 item, Cassileth Information Styles Questionnaire 5‐point Likert scale | ‐ | 24 hours post‐consultation, 3 weeks post‐consultation | |
| ‐ | ‐ | X Discordance between prognosis ratings Researcher‐administered questionnaire/interview; 7‐point scale; ‘discordance’ defined as difference of ≥ 2 between category ratings | Shortly after audio‐recorded consultation | |
| ‐ | X Preferences for amount and type of information Self‐administered, Cassileth Information Styles Questionnaire; scores subtracted from baseline preference scores; differences expressed dichotomously (preferences met or exceeded score > 0; unmet < 0) | ‐ | 1 month | |
| X: outcome assessed; ‐: outcome not reported. The study by Lautrette 2007 reported outcomes related to knowledge (ratings of time allocated for information provision, information clarity, and information seeking by family). As these were assessed at 90 days after the death of the patient, following delivery of a one‐off family conference intervention, we judged that these outcomes were too far removed in time from the intervention to be meaningful. Data were not extracted for analysis from this study, but data are provided in Additional table 9, for transparency. | ||||
| Primary outcome: evaluation of the communication
| ||||||
|---|---|---|---|---|---|---|
| Study ID | Quality of communication | Satisfaction with consultation (communication) | Patient‐centred communication | Preferences for involvement | Goal‐consistent care | Patient‐physician relationship |
| ‐ | ‐ | X Person‐centred approach to care Care and Activities and Interpersonal Relationships and Interactions domain of Person‐Centred Environment and Care Assessment Tool; 18 items, each rated 0 (not at all) to 3 (all of the time); rated by observation, resident and family reports and documentation No data | ‐ | ‐ | ‐ | |
|
| X Quality of Communication questionnaire, (scored 0 to 100, higher better); 2 weeks | ‐ | ‐ | ‐ | ‐ | ‐
|
|
| X Quality of Communication scale, timing, scoring unclear No data | ‐ | ‐ | ‐ | X No goals met. Life Priorities Survey (patients), Family Perceptions Survey (surrogates). Baseline, 2‐monthly. Scored by matching patient final Life Priorities survey (within 3 months of death) to Family Perceptions; score 0 to 3 goals met | X Therapeutic Alliance. Human Connection Scale. Baseline, 14, 24 weeks. Total score 7 to 28; higher better |
|
| ‐ | X Questionnaire, Roter and Korsch; 25‐item scale (25 to 125); 24 hours and 3 weeks post‐consultation; higher score better | ‐ | X Actual versus preferred involvement in consultation Questionnaire 24 hours post‐consultation 5‐item rating scale (ranging from doctor leads decisions to patient leads decisions) | ‐ | ‐ |
|
| ‐ | ‐ | X Composite patient‐centred (patient‐doctor) communication Composite of 4 communication measures; coded consultation; first visit after coaching session (intervention) or study entry (control) | ‐ | X Decision regret (family) Modified decision regret scale; 8 items, 2 months post‐mortem | X Patient‐physician relationship Human Connection Scale, Health Care Communication Questionnaire, Perceived Efficacy in Patient‐Physician Interactions scale; 2 to 4 days after audio‐recorded consultation, then quarterly |
|
| ‐ | ‐ | ‐ | X Control preferences (doctor/patient +/‐ carer involvement in decisions) Self‐reported questionnaire; baseline and 1 month; Degner Control Preferences Scale; scores subtracted from baseline, differences dichotomised, preferences met/exceeded score > 0; unmet < 0 | ‐ | X Patient Communication Self‐Efficacy Self‐reported questionnaire; baseline and 1 month; Perceived Efficacy in Patient/Physician Interactions Scale |
| X: outcome assessed; ‐: outcome not reported. Satisfaction with the intervention was reported by both Clayton 2007 and Walczak 2017, but only for the intervention arm. This was therefore not extracted and reported in the review. | ||||||
| Primary outcome: discussions of EoL/EoL care e.g. frequency, length, type, participants | |||||||
|---|---|---|---|---|---|---|---|
| Study ID | Discussion with clinicians | Discussion with surrogates | Documented discussion EoL care planning | Number of questions in consultation | Conversation duration | Communication content | Conversation timing |
| ‐ | ‐ | ‐ | ‐ | X Facilitated case conference intervention duration (non‐comparative) | ‐ | ‐ | |
|
| X (at last visit) Self‐reported questionnaire; 2 weeks after consultation | X (since last visit) Self‐reported questionnaire; 2 weeks after consultation | ‐
| ‐ | ‐ | ‐ | ‐ |
|
| ‐ | ‐ | X Conversation number/patient EMR review; after death | ‐ | X Physician report, post‐consultation (non‐comparative) | X Conversation content/quality SIC domains/patient EMR review; after death; coded 0 to 4 on number of domains discussed and documented (≥ 1 SIC, discussion about: values/goals, prognosis/illness understanding, EoL care, life‐sustaining treatment preferences) | X Timing of first documented SIC before death EMR review; after death |
|
| ‐ | ‐ | ‐
| X Questions, concerns, items tallied across 9 QPL categories; post‐consultation; coded 112 questions QPL grouped in 9 categories; 85 issues covered by QPL | ‐ | ‐ | ‐ |
|
| ‐ | ‐ | ‐
| ‐ | X At time of family conference with ICU staff | ‐ | ‐ |
|
| ‐ | ‐ | ‐ | X Number of questions, cues from patients, carers Coded audio‐recorded consultation; 1 week post‐CSP session; coding to identify number of direct questions, cues plus those on prognosis, EoL care, future care options, general issues | ‐ | ‐ | ‐ |
| X: outcome assessed; ‐: outcome not reported. CSP: communication support programme; EMR: electronic medical record; EoL: end of life; ICU: intensive care unit; QPL: question prompt list; SIC: serious illness conversation. | |||||||
| Secondary outcomes: health practitioner knowledge and understanding; evaluation of communication, communication encounter or preparedness to communicate | |||
|---|---|---|---|
| Study ID | Evaluation of communication | Knowledge and understanding | Preparedness to communicate |
| ‐ | Staff attitudes to, knowledge of providing palliative/EoL care PCPC assessed after training; other facility staff assessed after training by PCPC; Palliative Care for Advanced Dementia tool, 35 items (qPAD) No data | Staff confidence in providing palliative/EoL care PCPC assessed after training; other facility staff assessed after training by PCPC; Palliative Care for Advanced Dementia tool, 35 items (qPAD) No data | |
| ‐ | ‐ | Uptake and effectiveness of clinician training Use of conversation tool | |
| Physician satisfaction with consultation 24 hours and 3 weeks post‐consultation | ‐ | ‐ | |
| Bernacki 2019 also reported measures of clinician uptake and effectiveness of training to use their tool, and use of the tool. These were judged as measures related to implementation, rather than effects, of the intervention and data were therefore not analysed in this review. EoL: end of life; PCPC: palliative care planning co‐ordinator. | |||
| Secondary outcomes: health systems impacts (quality of EoL care, ratings of concordance with patient preferences for EoL care) | |||
|---|---|---|---|
| Study ID | Quality of EoL care (nurse rated) | Quality of EoL care (patient/family rated) | Ratings of concordance with patient preferences |
|
| X Nurse ratings of · CAD‐EOLD (higher scores = more comfort) · SM‐EOLD (higher scores = lower symptom frequency) Face‐to‐face/telephone interview; as soon as possible after patient death | X Family ratings of · CAD‐EOLD last 7 days life (higher scores = more comfort) · SM‐EOLD last 90 days of life (higher scores = lower symptom frequency) · SWC‐EOLD last 90 days of life (higher scores = more satisfied) Face‐to‐face/telephone interview; 4 to 6 weeks after patient death | ‐ |
|
| ‐ | ‐ | X PEACE scale, 2 subscales: · Peaceful Acceptance of Illness (acceptance of diagnosis, inner calm, feelings of being well‐loved); 5 questions, total score 5 to 20 · Struggle with Illness (feelings of upset, worry, anger, etc.), 7 questions, total score 7 to 28 Baseline, 2‐monthly |
|
| ‐
| X Caregiver evaluation of quality of EoL care 2 months post‐mortem, 6 items | ‐ |
|
| ‐ | X Surrogates' ratings of · SM‐EOLD last 90 days of life (higher scores = lower symptom frequency) · SWC‐EOLD last 90 days of life (higher scores = more satisfied) · Satisfaction with care (higher scores = better care) Interview, questionnaire; baseline, 3 and 6 months | ‐ |
| X: outcome assessed; ‐: outcome not reported. CAD‐EOLD: Comfort Assessment In Dying with Dementia; EoL: end of life; SM‐EOLD: Symptom Management at the EoL in Dementia; SWC‐EOLD: Satisfaction with Care at EoL in Dementia. | |||
| Secondary outcomes: health systems impacts (costs of care, hospital (re)admissions) | ||||||
|---|---|---|---|---|---|---|
| Study ID | Costs | Hospital (re)admissions | Consultation length | Treatments and hospice use | Timing (longest follow‐up) | Scale, scoring |
| X Training, CC and routine healthcare costs | ‐ | ‐ | ‐ | ‐ | Cost utility (benefit estimated as QALYs). QoL for economic analyses assessed by nurse‐rated EQ‐5D‐5L No data | |
| ‐ | ‐ | X | ‐ | ‐ | ‐ | |
| ‐ | ‐ | ‐ | X | ‐ | Treatments and hospice use in last month of life; medical records, composite score of 3 indicators of aggressive treatment in last 30 days of life: chemotherapy, potentially burdensome interventions, ED/hospital admission) and hospice utilization | |
| ‐ | ‐ | X | ‐ | ‐ | ||
| X: outcome assessed; ‐: outcome not reported. Data from Epstein 2017 (composite of treatments and hospice use in last month of life) was judged as clinical, rather than primarily related to communication; data were not extracted on this outcome for the review. CC: case conference; ED: emergency department; QoL: quality of life; QALY: quality‐adjusted life‐year. | ||||||
| Secondary outcomes: patient/family member/carer quality of life | |||
|---|---|---|---|
| Study ID | Patient quality of life | Timing (longest follow‐up) | Scale, scoring |
| X | 3‐monthly | Quality of life in Late‐stage Dementia (QUALID); 11‐item scale. Nurse‐rated No data | |
| X | Unclear | SF‐12 V2 health survey (QoL and general physical health function) No data | |
| X | 3‐monthly to 3 years | Composite QoL score as average of 5 z‐scored subscales: McGillQoL scale single item, McGill Psychological Well‐Being subscale, McGill Existential Well‐Being subscale, FACT‐G Physical Functioning subscale, FACT‐G Social Functioning subscale | |
| X | 1 month | Health‐related QoL (FACT‐G) | |
| X: outcome assessed. QoL: quality of life. | |||
| Study ID | Outcome category | Outcomes reported | Results | Assessment method and timing |
|---|---|---|---|---|
|
| Knowledge and understanding | Time for information | Sufficient time Intervention 51/56 (91%) Usual care 45/52 (87%) | Surrogate telephone interview; 90 days after death of patient |
| Clarity of information | Information was clear Intervention 52/56 (93%) Usual care 45/52 (87%) | Surrogate telephone interview; 90 days after death of patient | ||
| Additional information requested by family | Additional information was requested Intervention 17/56 (30%) Usual care 24/52 (46%) | Surrogate telephone interview; 90 days after death of patient | ||
| Anxiety (moderate to severe symptoms; GAD‐7) | 10.4% (intervention) versus 4.2% (usual care) | 24 weeks post‐baseline (approximately 12 weeks post‐intervention) | ||
| Anxiety (STAI) | 38.7 (intervention) versus 37.5 (usual care) | 3 weeks post‐intervention | ||
| Anxiety (HADS, score > 8) | 44.6% (intervention) versus 67.3% (usual care) | 90 days after death of patient | ||
| Data at longest time point are reported unless otherwise indicated. HADS: Hospital Anxiety and Depression Scales; STAI: State‐Trait Anxiety Inventory. | ||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 Doctor‐patient relationship Show forest plot | 3 | 238 | Std. Mean Difference (IV, Random, 95% CI) | 0.23 [‐0.06, 0.51] |
| 1.2 Discussion of EoL care planning Show forest plot | 2 | 537 | Risk Ratio (M‐H, Random, 95% CI) | 1.96 [1.61, 2.39] |
| 1.3 Patient questions in consultation Show forest plot | 2 | 249 | Mean Difference (IV, Random, 95% CI) | 1.58 [‐1.82, 4.98] |
| 1.4 Family‐rated symptom management (SM‐EOLD) Show forest plot | 2 | 212 | Mean Difference (IV, Random, 95% CI) | ‐1.98 [‐4.38, 0.43] |
| 1.5 Family‐rated satisfaction with care at EoL (SWC‐EOLD) Show forest plot | 2 | 212 | Mean Difference (IV, Random, 95% CI) | 0.44 [‐0.99, 1.87] |
