Diversity of Parent Emotions and Physician Responses During End-of-Life Conversations

This study was part of a larger research project examining end-of-life communication with families in ICUs (FamICom).^30–33  Before data collection, all physicians and nurses from the participating medical centers received oral and written project information. All but 1 NICU nurse gave their consent to participate. The inclusion period lasted from April 2018 to December 2019. For the current study, we analyzed the audio recordings of 49 conversations between physicians, nurses, and parents of children admitted to the NICUs and PICUs of 3 Dutch university medical centers, using a qualitative inductive approach.³⁴  Data were coded and analyzed with the qualitative software tool MAXQDA 2020,^35,36  and the Standards for Reporting Qualitative Research were followed.³⁷ 

The physician in charge informed the researchers when during the child’s care doubts had arisen within the medical team or by the parents whether continuing LST was still in the child’s best interest. From that moment on, parents were eligible to participate. A member of the research team or the attending physician then approached eligible parents to provide more information about the project and to ask whether they wished to participate. Parents could withdraw their consent to participate at any moment. All parents who were approached wished to participate, except for 1 family. From the moment of inclusion, audio recordings were made of all physician-parent conversations until a final decision was reached to either continue or discontinue LST. One university’s institutional review board approved the study protocol on behalf of all centers (W17_475 #17.548).

The dataset of the FamICom-project consisted of audio recordings of HCP-parent conversations concerning 36 children (NICU [n = 19], PICU [n = 17]). Purposive sampling was used to strive for maximum variation regarding the children’s age, diagnosis, disease progression, and course of treatment, and to parents’ ethnicity, level of education, and religion.

Audio recordings were transcribed verbatim, including nonverbal vocalizations (eg, crying, sobbing) and verbal expressions. Anonymized transcripts were inductively coded and analyzed until saturation was reached.³⁸  Details of the steps of coding and analysis are provided in Supplemental Fig 2.

A researcher with experience in health communication and qualitative research methods (SP), a researcher in the field of health communication and patient-provider interactions (AJL), and a behavioral scientist, ethics consultant, and researcher who is an expert in the field of family-centered care, end-of-life conversations, and qualitative research methods (MAdV) carefully read and reread the transcripts, while simultaneously listening to the audio recordings.³⁹  The 3 researchers decided to code inductively because previously described lists and codebooks did not fully capture the range of emotions and responses in the current study.^39–42  Because nurses rarely spoke in the audio-recorded conversations, the researchers solely focused on physician-parent interaction. The researchers coded which emotions parents expressed in each turn of talking and physicians’ direct responses to these expressions. The researchers also coded whether parents explicitly or implicitly expressed their emotions. In accordance with definitions in the literature, the researchers identified emotional expressions to be explicit if they directly referred to an emotional state (eg, “I am angry”). The researchers identified emotional expressions to be implicit if they indirectly referred to an underlying emotion (eg, “Come on, that’s why you are a physician, right?”).^43,44 

Parents could express multiple types of emotions in 1 single turn of talking. Similarly, physicians could use a combination of responses to these various emotions. The researchers, therefore, coded all expressed emotions and all physician responses to these emotions. As such, the total number of parents’ emotional expressions does not equal the total number of physicians’ responses. A preliminary codebook was developed iteratively. The applied codes and descriptions of these codes were extensively discussed by the researchers. Coding continued until saturation was reached.³⁸  This concretely meant that the researchers stopped the inclusion of new transcripts and the corresponding audio recordings from the moment on that they did not identify any new parental emotions nor any new physician responses. The applied codes were then elaborately analyzed, discussed, and reflected on during 15 discussion rounds with the research group in different constellations to identify patterns of overarching emotions and responses.

On the basis of the “Phases of Illness” classification within palliative care,^45,46  we categorized all conversations into 3 phases of illness and decision-making to structure the conversations and grasp the emotional context: the initial phase (ie, the child’s condition was unstable, and additional diagnostics and care planning were urgent), the middle phase (ie, the child’s condition deteriorated and treatment options depleted), and the last phase (ie, the child’s death was expected within days; Supplemental Table 4).

Forty-nine conversations (NICU [n = 29]; PICU [n = 20]) regarding 12 patient cases (NICU [n = 6]; PICU [n = 6]) were included before we reached saturation. In all but 1 PICU case, physicians and parents had multiple conversations (range 2–10) (Table 1). In total, 12 mothers, 10 fathers, and 46 physicians participated in the conversations. In all 49 conversations, parents expressed at least 1 emotion implicitly (range 2–28). We identified a total number of 499 implicit expressions of emotions. Parents expressed their emotions less frequently explicitly. Thirty-seven conversations (76%) contained at least 1 explicit emotional expression (range 1–16). Overall, we identified 186 explicit expressions of emotions. In 23 conversations, parents expressed their emotions in nonverbal vocalizations, particularly by crying.

Across all 49 analyzed conversations, we identified 22 distinct emotions expressed by parents. These emotions (henceforth: sub-emotions) were categorized into 7 overarching emotions: anxiety, anger, devotion, grief, relief, hope, and guilt (Table 2).

Anxiety contained the broadest array of sub-emotions, which parents mostly uttered implicitly. All conversations were fraught with such expressions of anxiety.

During initial-phase conversations, parents’ expressions of anxiety concerned various topics, ranging from the child’s uncertain diagnosis to worries about diagnostic tests, the course of treatment, and its consequences. In middle-phase conversations, parents mainly expressed anxiety regarding the treatment plan and associated risks. They also voiced their feelings of powerlessness and insecurity as they felt unable to take care of their children. In last-phase conversations, parents expressed their despair and fear for their child’s wellbeing during the dying process.

Parents expressed their anger verbally and in nonverbal vocalizations (eg, slamming their fists on a table). In initial-phase conversations, parents uttered their anger about how their situation deviated from their expectations of parenting. They also expressed their jealousy, for instance, of parents with healthy babies. In middle-phase conversations, parents’ anger focused on the timing, amount, and quality of the provided information. Parents also expressed anger about the perceived inadequacy of their child’s care and the perceived incompetence of certain HCPs. Additionally, parents shared their frustration about the situation or the lack of understanding from others. During last-phase conversations, parents rarely expressed anger.

In all cases, parents expressed their devotion to their child (eg, “when you see his heart dip, it is like my own heart is pounding”). In initial-phase conversations, parents stressed their strong bond and desire to care for their child. In middle-phase conversations, parents either implicitly or explicitly expressed their strong urge to protect their child. Some parents did so by underlining their wish that everything should be done to keep their child alive, whereas other parents expressed that their child should not be treated “at all costs.” In last-phase conversations, parents stated their desire to be close to their child and do everything to help and comfort their child, no matter how unbearable this would be for themselves.

Parents expressed their grief verbally and in nonverbal vocalizations, by crying and sobbing. In initial-phase conversations, parents uttered sorrow in response to the negative news they had just heard and about the fact “of being in this situation.” In middle-phase conversations, parents expressed grief in response to the bad news provided by the physician. Notably, parents showed fewer expressions of grief in this middle phase compared with the other phases. In last-phase conversations, parents expressed grief about the impending loss of their child. They uttered grief at witnessing their child suffer and realizing that their child would never come home again.

Parents uttered relief verbally and in nonverbal vocalizations by laughing or sighing. During initial-phase conversations, parents expressed relief when the physician clearly explained that the child’s situation was not the parents’ fault. In middle-phase conversations, parents expressed relief when physicians provided relatively “good” news, for instance, that the child’s condition had not further deteriorated. Occasionally, parents voiced their relief by thanking their physician for the provided information or the care given to their child. In the few instances in which parents disclosed relief during last-phase conversations, they stated to be grateful for the “silver linings” (eg, that they could hold their child during the dying process).

Parents expressed their hopefulness explicitly and implicitly. The latter occurred in 2 ways: (1) asking loaded questions with hopeful implications (eg, “if he deteriorates, you can help him, right!?) or (2) sketching a positive scenario (eg, “let’s assume that it is going to be alright”).

In initial-phase conversations, parents’ expressions of hopefulness primarily concerned the hope that the child’s condition would improve. In middle-phase conversations, parents redirected their hope to wishing that their child’s condition would not further deteriorate. Parents also more explicitly uttered the hope that the cause of the child’s illness would be found. In last-phase conversations, parents’ expressions shifted to the hope that their child would die peacefully. Parents also uttered the hope that their child would “choose” the moment to pass away so that they themselves would not have to make this decision.

Parents expressed feelings of guilt mostly in an implicit way. In conversations in the initial phase, parents uttered feelings of guilt about whether it was their fault that their child had become so ill or whether they should have acted differently to prevent their child’s situation from becoming so severe. In middle-phase conversations, parents stated that “they could never forgive themselves” if their child’s remaining quality of life would be extremely poor, especially in the case in which they had decided to continue their child’s life-sustaining treatment. At the same time, parents expressed how guilty they would feel if they had not done everything to let their child survive. Some parents explicitly verbalized how “ambivalent this all felt.”

In last-phase conversations, when deciding on the discontinuation of treatment, parents worded their feelings of guilt as “giving up” on their child.

We identified 13 types of physician responses to parents’ emotions, ranging from cognition-oriented to affect-oriented. Table 3 contains all responses and specifies whether these responses provided or reduced the opportunity for parents to further elaborate on their emotions. Below, we describe prevalent or notable patterns in physicians’ responses to parents’ emotions and parents’ emotional reactions during a conversation (illustrative quotes presented in Table 3).

We noticed that parents did not further disclose and elaborate on their primary expressions of often implicit anxiety if physicians responded in a cognition-oriented way to these expressions (Table 3, Q2.1 and Q6.1). In the remainder of the conversation, parents started to stress their strong devotion to their child, also in combination with anger (Table 3, Q5.2). In response to the changed pallet of emotions, physicians continued their information provision rather than identifying parents’ covert anxiety and then inviting them to elaborate on these anxieties. By contrast, when physicians immediately responded to parents’ implicit anxiety by explicating these expressions (Table 3, Q10.1) and further exploring it (Table 3, Q11.1), parents subsequently clearly worded what they were afraid of. Physicians were then able to provide detailed and targeted explanations that addressed these specific anxieties.

When the physician continued providing explanations or started to defend the chosen course of action in response to parents’ expressions of anger, this seemed to intensify parents’ distrust toward the medical team and sense of protectiveness toward their child (Table 3, Q1.1). Such cognition-oriented responses particularly occurred when parents expressed anger about a topic for which they held the physician or other HCPs accountable, like the withholding of important medical information. Conversely, affect-oriented responses seemed to occur when parents’ anger concerned a topic that could be perceived as being “beyond anybody’s control” (eg, a pessimistic prognosis). When physicians responded in an affect-oriented way to parents’ expressions of anger (Table 3, Q10.2, Q10.3, and Q11.2), it appeared that parents’ anger dissipated, and instead gave way to expressions of anxiety and grief in the remainder of the conversation.

When physicians responded in a cognition-oriented way to parents’ expressions of devotion, for instance, by arguing or rephrasing (Table 3, Q3.2, Q4.1, and Q4.2), parents re-expressed their devotion or started to express mistrust in the medical team. We observed a similar pattern when physicians briefly acknowledged parents’ expressions of devotion (ie, affect-oriented), for example, by stating “I understand” or “I know.” Subsequently, parents also started to stress what they really wanted for their child and to emphasize their frustration (Table 3, Q9.1). In turn, these combined expressions of devotion and anger appeared to elicit a defensive response from physicians. By contrast, when physicians acknowledged parents’ expressions of devotion, for instance, by wording their respect for it or attempting to understand parents’ viewpoints, parents reacted differently (Table 3, Q9.2). Then, they started to disclose their fears, while at the same time expressing their gratitude for the medical team’s efforts in doing everything possible to keep their child alive.

When parents expressed grief, we noticed that physicians primarily responded in an affect-oriented way (Table 3, Q12.1, Q12.2 and Q13.1). This especially occurred in conversations when the child’s death was imminent. After such affect-oriented responses, parents further expressed their grief and disclosed their anxiety. In turn, physicians responded in an affect- and cognition-oriented way, for example, by showing compassion (Table 3, Q13.2) and providing detailed information to prepare parents for what would await them and address their fears (Table 3, Q7.2). In the rest of the conversation, parents expressed a variety of emotions ranging from anxiety to relief.

When physicians solely responded in a cognition-oriented way to parents’ expressions of hope (Table 3, Q3.1 and Q5.1), these expressions kept reappearing throughout the remainder of the conversation. On the contrary, when physicians responded in both an affect- and cognition-oriented way, for instance, by acknowledging parents’ hope while also managing their expectations (Table 3, Q7.1), parents did not further dwell on their hopes, but seemingly felt the opportunity to express their underlying anxiety.

We noticed a different pattern when physicians briefly responded in an affect-oriented way to parents’ expressions of guilt and did not provide any explanations (Table 3, Q8.1). In these instances, parents’ expressions of guilt appeared to persist during the remainder of the conversation. By contrast, in 1 case the physician immediately clearly explained that the parents were not responsible and could not have prevented the severity of their child’s situation (Table 3, Q5.3). In this case, the parents did not express any further guilt throughout the rest of the conversation and instead expressed a sense of peace with their situation.

Our study offers detailed and real-life insights into the various emotions that parents expressed during conversations in the NICU and PICU when faced with doubts about whether continuing LST was in their child’s best interest. Physicians mainly responded to parents’ emotions in a cognition-oriented way. These types of responses appeared to reduce the opportunity for parents to further elaborate on their emotions. In turn, the parents’ emotions seemed to persist or intensify in the remainder of the conversation. A complementary approach, in which physicians first provide parents the opportunity to disclose their emotions, followed by a cognition-oriented response, may offer parents both emotional and informational support.

The observed variety of parents’ emotions, especially within the overarching emotions of anxiety and anger, was more diverse than previously reported in studies in the NICU and PICU,^{1,2,4,12,47,48}  as well as in other medical contexts.^6,13,17,49  A thorough insight into the wide array of emotions that parents may express during end-of-life conversations is especially important because these emotions also reflect how parents perceive and evaluate their child’s situation and care. Moreover, these emotions may influence parents’ treatment preferences and the role they wish to have in the decision-making process.^15,50,51  More knowledge about helpful and less helpful responses to parents’ emotions, either implicitly or explicitly expressed, enables physicians to choose the most appropriate approach, for example, taking sufficient time to empathically explore what parents concretely fear. This information will add to a shared understanding and to mutual trust between physicians and parents.^50,52  When physicians are less attentive to parents’ diverse emotions, they may easily miss opportunities to deepen the conversation. By effect, valuable information likely remains unknown, and this may even hinder the making of a balanced treatment decision.^2,15,52 

We found that all conversations were fraught with parents’ often implicit expressions of anxiety. When physicians responded to parents’ emotions in an affect-oriented way, particularly by exploring, parents started to disclose this anxiety, irrespective of the type of emotion they had expressed before. Anxiety seemingly becomes apparent when parents are given the opportunity to elaborate on it. If not, other emotions tend to mask it.

Physicians particularly seemed to respond in an affect-oriented way when parents’ emotions were more “tangible” (eg, grief when the child’s death was imminent), “explicit” (eg, clearly stating they feel anxious), or being beyond anybody’s control (eg, anger concerning the pessimistic prognosis). More often, physicians responded in a cognition-oriented way to parents’ emotions, which is in line with previous studies across various medical settings.^{8,10,13,53–57}  After such cognition-oriented responses, parents’ anxiety remained under the surface. Instead, expressions of anger and protectiveness predominated the remainder of the conversation. This seemed to make it even more difficult for physicians to uncover parents’ specific fears. Previous studies reveal that physicians find it hard to adequately respond to emotions and instead choose to focus on providing medical information.^{26,50,55,58–62}  Our findings reveal that it might be more constructive to explicitly explore what parents are specifically anxious, angry, or protective about before providing medical information. This enables physicians to first uncover hidden anxieties and then provide tailored information that aligns with parents’ fears and feelings. According to October et al,¹⁸  unburied empathy (eg, asking open-ended questions to explore emotions instead of using medical terminology or “but” statements) can be an effective way to address and explore families’ emotions, thus making them feel heard. This approach also allows physicians to understand the values that are most important to the family, which can increase feelings of trust and control, improve the physician-parent relationship, and can lead to better decision-making.^18,63 

Additionally, former studies reveal that when physicians “provide space” to patients to express their emotions, this helps physicians to subsequently choose the most appropriate approach to respond to these emotions.^10,55,57,60  In line with these outcomes, we found that when physicians provided parents the opportunity to further elaborate on their emotions, most parents subsequently expressed their emotions more explicitly. This indeed seemed to help physicians to choose adequate responses.

Notably, some responses that were previously classified in the literature as providing space (eg, reassuring and acknowledging⁴⁰ ) sometimes appeared to reduce space in our study. In these instances, emotions intensified, for instance, into anger, or parents kept repeating the same emotion throughout the conversation, for instance, feelings of guilt. This may partly be explained by how physicians express their reassurance or acknowledgment. If they do so too briefly or prematurely and without really listening to parents’ perspectives, such responses may be counterproductive.^18,64  When emotional support is not accompanied by additional explanations, this may also be counterproductive.⁶⁵  Thus, it may be important that physicians primarily give parents the openness and trust to express what is really on their minds. Based on this trust, there is room for physicians to give tailored information, ultimately providing parents with support in both affective and cognitive ways.^65–67 

Although presenting novel and in-depth insights, our study is subject to several limitations. The most significant limitation is that we do not know for sure that the emotions we identified were indeed the emotions felt by parents at that point in time, especially when it comes to the emotions that parents expressed in an implicit way. Although we cannot be certain that we have correctly identified all emotions, our study may well serve as the first cornerstone for future research in this specific field of communication. Additionally, we were unable to investigate all nonverbal communication because we only audio-recorded HCP-parent conversations to minimize intrusion. Besides, we could not audio-record informal bedside conversations because of privacy regulations. Although we have reached saturation regarding the emotions that parents expressed in our dataset and regarding physicians’ responses, future research regarding end-of-life conversations in other ICUs may well reveal other emotions and responses and, by effect, identify additional patterns in how these emotions evolve under the influence of HCPs’ responses. Particularly, research in other countries may reveal new insights and shed light on cross-cultural differences. In addition, we do not know how parents themselves would define their emotions, nor do we know how they experienced physicians’ responses to their emotions. We also do not know how physicians experienced the different types of emotions that parents expressed, nor do we know how they evaluated their own responses in retrospect. In that respect, our exploratory research can be seen as the first step in this important but also complex field of research. To both validate and expand our findings, it would be necessary to combine the analysis of video-recorded conversations with retrospective interviews with the families, physicians, and nurses who participated in these conversations. Exploring how HCP’s responses affect parents’ emotions not only during but also after conversations is also a worthy subject for future studies.

The main strength of our study is that we meticulously analyzed real-life conversations. This enabled us to present a detailed overview of parents’ actual emotional expressions and physicians’ responses during complex and emotionally charged conversations in the NICU and PICU. (Figure 1. Practice Recommendations)

Parents expressed a broad diversity of emotions in all conversations, mostly in an implicit way. In responding to these emotions, physicians appeared to reduce the opportunity for parents to elaborate on their emotions by predominantly responding in a cognition-oriented way. Subsequently, the parents’ emotions seemed to persist or intensify. Affectively exploring and explicating parents’ emotions appeared an effective response in uncovering complex emotions and hidden anxieties. If presented in a sincere way, a complementary approach of affect- and cognition-oriented responses could provide parents the opportunity to elaborate on their emotions and offer parents emotional and informational support during turmoil times.

We are extremely grateful for the trust of the participating families, physicians, and nurses. We would also like to thank Sandra Dijkstra, Annelies Olthuis, Joke Dunk, Odile Frauenfelder, Maartje Harmelink, and Joyce Lamerichs for their valuable assistance in the data collection phase of this study. In addition, we thank Vicky Lehmann for correcting the English manuscript.

HCP

health care provider

LST

life-sustaining treatment