How doctors manage conflicts with families of critically ill patients during conversations about end-of-life decisions in neonatal, pediatric, and adult intensive care

This qualitative exploratory study was part of a larger research project (FamICom) on communication about end-of-life decisions with families in IC [21]. Data were derived from audio-recordings of family conferences (henceforth: conversations) in the NICU, PICU, and ICU of the Amsterdam University Medical Centre.

Families of 36 patients and 71 doctors participated. Table 1 lists their characteristics. Maximum variation was sought regarding patients’ age, sex, diagnosis, disease progression and course of treatment, and families’ ethnic background, level of education, and religious beliefs. ‘Families’ refers to family members or close friends who attended the conversations.

Prior to data collection, all IC-doctors and IC-nurses received oral and written information about the study and were asked for their consent to participate. All doctors and all but one nurse gave this consent.

The inclusion period lasted from April 2018 to December 2019. Families were eligible to participate from the moment that doubts arose whether continuing LST was still in the patient’s best interest. The attending doctor or nurse introduced the study to eligible families. Interested families were further informed and asked for their oral and written consent by a member of the research team or the attending doctor. All but one family decided to participate.

The conversations were almost always planned beforehand on initiative of the doctor and almost never on request of families. Only incidentally conversations took place because patients’ situation acutely deteriorated. From the moment of inclusion, all conversations were audio-recorded by the attending doctor until a final decision was made. All conversations took place seated around a table in one of the conference rooms on the unit. At least one nurse was present during most of the conversations. However, due to nurses’ minimal (verbal) engagement in these conversations, this study focuses solely on the communication between doctors and families.

The audio-recordings were transcribed verbatim and anonymized. We then coded and analyzed our data, thereby using a deductive approach. This process consisted of four phases, as illustrated in Fig. 1. Coding and analysis of the transcripts were performed with MaxQDA 2020.

The Amsterdam UMC institutional review board waived approval of this study (W17_475 # 17.548). Informed consent was acquired from one representative on behalf of the whole family. Consent could be withdrawn at any time.

We identified four factors, either context- or family-related, that complicated and deepened conflicts that arose during conversations. First, uncertainty regarding patients’ diagnosis or prognosis appeared to reinforce conflicts. We noticed that when high levels of uncertainty were present, families appeared to be reluctant to follow or accept decisions proposed by the doctors.

Second, conflicts appeared to intensify if families became highly emotional. This always concerned the expression of negative emotions like fear, guilt, anger, distrust, or hostility. These emotions were expressed explicitly or implicitly, for example, by a changed tone of voice or snorting. Often it remained unclear whether conflicts triggered these emotions or vice versa.

Third, limited health literacy of families added to the complexity of conflicts. We observed that several families had a hard time understanding medical information provided by doctors, as became clear from their inability to summarize or to answer questions regarding this information. We noticed that in several conversations families’ misinterpretation of information coincided with their disagreement with the proposed treatment decision.

Fourth, families’ burden of responsibility added another layer of complexity. Families occasionally provided cues, both implicit and explicit, that they felt disproportionately responsible for the treatment decisions that were made. For example, in one case in which a mother disagreed with the doctor’s proposal to withdraw LST, this conflict was deepened by her explicit assumption to be solely and ultimately responsible for the decision to let her child pass away. In another case, a mother underlined that she felt highly burdened by her feelings of responsibility. At a later point in the conversation, she added that she could not agree with the decision to withdraw her child’s LST because of her religious convictions. This was the only conversation in which religious convictions played a role in the arising and deepening of a conflict.

Complicating factors often co-occurred. For example, diagnostic or prognostic uncertainty seemed to reinforce the burden of responsibility that families experienced. Additionally, families’ burden of responsibility often co-occurred with strong expressions of doubt and guilt. Families’ limited health literacy was often accompanied by expressions of anger and frustration.

We identified four overarching strategies that doctors used to manage conflicts: content-oriented, empathic, moral, and process-oriented strategies. Table 4 provides an overview of these strategies, their sub-strategies, and illustrative quotes.

To manage conflicts, doctors predominantly used content-oriented strategies, i.e. strategies focusing on the provision of or a request for information. These strategies specifically concerned extensive clarifying and explaining. Doctors used empathic strategies to a lesser extent. Moral and process-oriented strategies were least often used.

In contrast to the ICU, doctors in the NICU and PICU more often used empathic strategies. Yet, in most instances, short empathic responses were followed by lengthy explanations and clarifications.

Moral strategies were evenly applied in the three IC settings. Interestingly, doctors never directly inquired about families’ moral values, but solely introduced their own moral standpoints. Moral strategies often co-occurred with the content-oriented sub-strategy arguing.

In all units, doctors occasionally used process-oriented strategies, especially postponement, often combined with a content-oriented or empathic strategy. For instance, one doctor proposed to postpone the decision, clarified that this was done to give the family more time, and then acknowledged how hard the situation had to be for the family.

Content-oriented strategies appeared to be effective in managing conflicts regarding one topic. If this was the case, doctors could easily identify and address disagreements on a rational level. In more complicated conflicts, an effective approach consisted of the acknowledgment of the complicating factor(s) in an empathic and understanding way. For example, if prognostic uncertainty played a prominent role, acknowledgment of this uncertainty and the resultant burden on families appeared to nip conflicts in the bud. In conversations in which families became increasingly emotional, it proved to be effective if doctors not only uttered an empathic remark, but also took the time to explore what families were going through and how this made them feel. In this way, doctors constructed a common ground for a content-oriented follow-up. In this follow-up, doctors not only gave additional information, but also verified families’ viewpoints by asking them to expand on them. However, doctors only occasionally applied this combination of empathic and content-oriented strategies. If doctors explicated their own viewpoints, which rarely occurred, this appeared to open up a dialogue about the viewpoints and emotions of both doctors and families. This often appeared to create a common ground with families, which resulted in the resolution of conflicts. When doctors, despite the use of empathic strategies, were unable to create this common ground and the conversation threatened to end in an impasse, it often proved helpful to postpone the decision and transfer this topic to the next conversation.

In a minority of conversations, doctors’ strategies appeared to be ineffective as indicated by the fact that conflicts kept reappearing throughout the conversation. These lingering conflicts predominantly occurred in the NICU and PICU as opposed to the ICU.

We identified two distinct patterns, both resulting in the perseverance of conflicts. In the first pattern, doctors did not acknowledge and respond to families’ cues indicating the presence of one or more complicating factors. For instance, several families clearly hinted that they felt burdened by the prevailing diagnostic or prognostic uncertainties and/or by their responsibility for the outcome of the decision to discontinue LST. Instead of acknowledging and addressing this dual burden, most doctors kept using content-oriented strategies, especially extensive and repeated explanations. By effect, conflicts persevered and even deepened.

In the second pattern, doctors first employed content-oriented strategies after conflicts arose and then, when conflicts lingered on, switched to a moral strategy. To illustrate, in several conversations doctors stated that they would not resuscitate the patient if his or her heart would suddenly stop, because “this was not what good healthcare providers should do”. This appeared to fuel the disagreements and transform them into full-blown conflicts. In another conversation, after a doctor had stressed that withdrawing LST was “what must be done in the child’s best interest”, the mother kept repeating, more and more desperately, that she would not allow him to kill her child.