Empathy Expression in Interpreted and Noninterpreted Care Conferences of Seriously Ill Children Free

This was a prospective, mixed methods cohort study analyzing transcripts of care conferences for seriously ill children at a single quaternary academic pediatric hospital. It was approved by the local institutional review board. Informed consent was obtained from patients’ families and clinicians.

Study participants included hospitalized patients of any age with a planned care conference, family members (including at least 1 legal guardian; hereafter referred to as “family” or “parent”) and any clinical team members participating in the conference (hereafter referred to as “clinician”). Any upcoming care conferences (same day to within 2 weeks) were identified through communication with unit-based social workers and palliative care team members. To focus on patients with acute serious illness, we included patients in the PICU, cardiac ICU, NICU, or in any inpatient unit with a palliative care consult. After screening for eligibility, study personnel consulted with the clinical team. Families were excluded if the clinical team felt they were too distressed for recruitment. Bedside nurses were then asked to introduce the study and ask families if they were willing to hear more. If families agreed, they were approached in-person or by phone by research team members unknown to them. If family provided informed consent to participate, research staff obtained informed consent from all participating clinical team members immediately before the conference. Participants were told that the study goal was to assess communication in care conferences. Participants were enrolled from January 2018 to January 2021.

Patient clinical characteristics were obtained from electronic health record review. Families completed surveys which included: demographics (sex, age, race, ethnicity, highest level of education), language for care, self-assessed ability to speak English using US Census categories (not at all, not well, well, or very well), and previous care conference experience (none, 1 or 2, 3 or more).

At the beginning of each care conference, a study team member set up a recording device, then waited outside of the room to stop recording once participants were finished. Audio recordings were transcribed verbatim. All interpretation was performed using trained professional interpreters. For staff and agency interpreters at the institution, local, state, and national certifications are accepted. Because the study focused on how clinicians respond to what they hear (whether through an interpreter or directly), only English portions were transcribed. Transcripts were not shared with study participants.

We conducted qualitative thematic analysis following Standards for Reporting Qualitative Research.³⁸  We adapted existing frameworks to develop our directed content analysis approach, including a priori coding of: family expression of empathic opportunities, clinician expression of empathy or missed opportunity to express empathy, and family response to clinicians. For family empathic opportunities, we adapted the Verona Coding Definitions of Emotional Sequences, a consensus-based system for coding patient expressions of emotional distress, which has been validated in several languages and clinical settings.^26,39  It includes:

1. concerns, or expressions that refer directly to feelings and emotions (“This feels daunting”); and

2. cues, or less explicit expressions of emotion (“I just want her to be healthy…I want you guys to fix it.”).^22,24,26 

We adapted a framework from October et al to assess clinician responses.¹⁹  This coding approach uses a clinically relevant, well-known mnemonic for empathetic responses: name, understand, respect, support, explore, silence (NURSE[S]).⁹  When clinicians responded to an empathic opportunity with a NURSE(S) statement, this was further coded as unburied (if the statement was followed by a pause of >5 seconds to allow family to respond), or buried (if the statement was immediately followed by medical talk). When a clinician did not use a NURSE(S) statement in response to an empathic opportunity, this was coded as a missed opportunity, which was further subcoded (Supplemental Table 7).^9,40  Family responses to clinicians were coded and categorized into themes:

1. alliance responses: families shared more about emotions, agreed with clinicians, expressed mourning, or expressed gratitude;

2. cognitive responses: families did not acknowledge emotional content; and

3. no responses: families were either interrupted, not given an opportunity to reply, or gave a brief reply such as “ok” or “got it.”

The primary coding team represented pediatric physicians with experience in critical care and bioethics (A.E.O.), palliative care (J.B., A.T.), health services and qualitative research (A.E.O., A.T. and J.B.), and a research assistant with training in health services and qualitative research (A.M.). Two coders (J.B. and A.T.) were present for some care conferences. They did not code any conferences for which they were present. Coders had the potential to code for colleagues with whom they worked, but all transcripts were anonymized. To improve validity, an advisory panel with representatives from the patient family advisory council (K.W.); patient navigation and interpretation services (B.F.); the hospital’s Center for Diversity and Health Equity (M.S.); and palliative care, oncology, and bioethics (A.R.R.) was involved in all steps of analysis, from codebook development to data analysis and interpretation.

The first 8 transcripts were coded as a team of 4, for group training in the codebook (Supplemental Table 7), refinement of codes, and consensus- establishment. Remaining transcripts were coded in teams of 2. We used variable coder pairings to improve consistency. Because we coded as a team, we did not complete testing for interrater reliability. Given the inherent subjectivity of defining an empathic opportunity, the coding team reviewed all challenging excerpts with the advisory panel, using a final “vote” in rare cases when consensus was not reached. Dedoose software was used for data analysis and interpretation.⁴¹ 

We summarized patient and family demographics, survey responses, family empathic opportunities, clinician responses, and family responses using counts and percentages. We assessed statistical significance of differences between clinician expressions in care conferences with language interpretation and in English, and differences in family responses to categories of clinician statements, using Pearson χ² tests. Quantitative analyses were performed with a 2-sided α of 0.05, using Stata version 16 (StataCorp., College Station, TX).

We identified 69 eligible patients, of which 29 were included. The remainder were not included because: study team was unable to reach family (n = 10), study team was unavailable to record (n = 9), or family declined because they were overwhelmed (n = 5), they did not want to be recorded (n = 10), or no reason given (n = 6). Two families did not complete demographic surveys; the remainder of their data were included.

Median patient age was 8 months, and 18 of 29 (62%) of patients were female (Table 1). Most care conferences (19 of 29, 66%) were for patients hospitalized >1 month. The most common diagnoses were genetic, cardiac, and pulmonary (Table 1). Most patients (17 of 29, 59%) were in an ICU. Of family participants, 22 of 27 (81%) were female, and 19 of 27 (70%) had previous care conference experience. A total of 11 of 29 (38%) required language interpretation, with 8 in Spanish, 2 in Vietnamese, and 1 in Somali. All interpretation was provided in-person, with 1 conference starting with virtual interpretation then switching a few minutes in. Care conferences ranged from 17 minutes to 110 minutes (mean 44 minutes, SD 0.70 for English; mean 63 minutes, SD 1.10 for interpreted care conferences). The mean number of clinical team members, including language interpreters, was 8 (range 3–13, SD 2.5) for English, and 7 (range 5–11, SD 2.3) for interpreted care conferences (Table 2). The median number of family members present was 2 (range 1–3) for both English and interpreted care conferences (Table 2). Some conferences included virtual clinician participants. Care conferences were typically led by the primary team (86%, Table 2). Palliative care was present at 65% of care conferences (Table 2).

Across 29 care conferences, we identified 210 empathic opportunities, with a median of 5 (range of 0–20) per care conference. There was no difference between interpreted and English care conferences (median 7, mean 8, SD 6.2, range 0–20 for English; median 5, mean 6, SD 4.7, range 1–15 for interpreted). Overt concerns (explicit emotional statements) made up 27% of empathic opportunities for interpreted conferences, compared with 15% for English conferences (P = .048) (Table 3). Of the cues that represented more subtle hints of emotion, verbal hints at a concern (34%) or neutral phrases with emotional overtones (21%) were most common. We captured 7 instances of positive emotion (3% of total empathic opportunities).

In response to 210 total empathic opportunities, clinicians expressed empathy (with NURSE[S]) 58% (122 of 210) of the time, and did so in an unburied fashion 38% (80 of 210) of the time (Table 4, Fig 1). Of all NURSE(S) statements, they were most likely to use a “respect” statement (“I think what I’m hearing are some really thoughtful and loving things for [child], that you really want her to have a happy and long life”), (32%, 39 of 122, of empathic expressions). They were least likely to use silence/continuers or “explore” statements (“Tell me what you are worried about”), (8%, 10 of 122, and 6%, 7 of 122, respectively). When clinicians missed empathic opportunities, they most often did so by sharing medical information (52%, 46 of 88, of all missed opportunities, Supplemental Table 8).

We found no differences in clinician empathic expression between interpreted and English conferences (61% in interpreted care conferences, 57% in English care conferences, P = .53, Table 4, Fig 1). Likewise, clinician expressions of unburied empathy were similar for interpreted and English care conferences (39% of all responses in interpreted care conferences, 38% in English care conferences, P = .88, Table 4, Fig 1). We found no difference in clinician empathy expression in response to concerns compared with cues (60% for concerns, 58% for cues, P = .79). Clinicians were slightly more likely to express unburied empathy when family expressed concerns compared with cues, but this did not reach statistical significance (75% for concerns, 63% for cues, P = .28).

We coded 80 family responses to unburied empathic statements, 42 responses to buried empathic statements, and 88 responses to missed empathic opportunities (Table 5). When clinicians expressed unburied empathy, families were more likely to express alliance compared with when clinicians expressed buried empathy or missed empathic opportunities (83%, 14%, 15%, respectively, P < .01, Fig 2). These findings were similar for care conferences that used language interpretation (85%, 13%, 19%) and those that did not (81%, 15%, 13%). Table 6 shows exemplars of exchanges that include family empathic opportunity, clinician response, and family response.

In this mixed method study of care conference communication for 29 seriously ill patients and family members, we found that clinicians frequently missed opportunities to empathize in a manner that allowed for deeper discussion. Contrary to our hypothesis, our findings were similar for care conferences using language interpretation and those in English. When clinicians acknowledged emotion and created space to respond, families were far more likely to respond with alliance (>80% of the time), including deepening or sharing more about their emotion, expressing gratitude, and expressing agreement. However, when clinicians buried empathy with more medical talk or missed opportunities to express empathy, families most often responded cognitively or not at all (86% for missed opportunities, 85% for buried empathy).

Our study is consistent with previous work, which suggests that, when clinicians display empathy, families share more about their feelings and goals, the clinician–family relationship is enhanced, and families report feeling more supported.^9,42,43  Our findings differ from a previous study examining adult care conferences, which showed that clinicians express support less frequently when language interpretation is used.¹⁸  There are few studies on interpreted care conference empathic expression (none in pediatrics), and the most recent adult study is from 2009. Thus, the difference we found may indicate improvements in communication with language interpretation over the past 12 years, as well as a potential difference between pediatric and adult settings.

It is encouraging that we did not find a difference in communication of empathy between patients who use language interpretation and those who speak English. One possible reason for this finding is that professional interpretation allows clinicians to recognize and respond to emotions as effectively as if there was no language difference. This finding is supported by previous work showing that strategies to improve communication for patients who use language interpretation led to improvements in care quality and patient satisfaction.^{14,15,20,44–49}  For example, improving access to and use of interpretation led to improved informed consent processes, decreased hospitalization lengths of stay, and decreased readmission rates.^{14,15,20,50–55}  In 1 systematic review, improved access to professional interpretation was associated with improved communication (errors and comprehension), utilization, clinical outcomes, and satisfaction with care.⁵⁶  This finding highlights the critical need for consistent use of professional interpretation. Furthermore, though in this study, empathic expression may not be a source of disparities when professional interpretation is present, we know that significant disparities in outcomes and communication still exist for families who use language interpretation, and further work is indicated to explore and improve them.^{10,11,13,14,17,57} 

In addition to highlighting how empathy expression positively impacts family responses in care conferences, our results suggest that burying clinicians’ empathic responses with more medical talk is as ineffective as missing the opportunity to express empathy altogether. This suggests that clinicians seeking to improve their communication and expression of empathy may need to focus on creating space for families to respond to empathic statements. As empathy continues to be better understood empirically, we can refine how we teach its expression with techniques such as silence.^52–54  Additionally, when clinicians miss empathic opportunities, unsurprisingly, they most commonly share medical information instead. Clinicians are trained to integrate and share medical information; focusing on medical details may be the most comfortable approach for many.⁵⁵  Understanding common ways clinicians miss opportunities to express effective empathy may help guide their approach.

This study has several important limitations. We included a small overall number of care conferences and a limited range of languages. A significant number of conferences were excluded because parents declined participation. It is possible that parents’ communication with the clinical team impacted their decision, which may have impacted our findings. Furthermore, we excluded families who were too distressed for recruitment, which may have impacted our findings. Anyone who attended the care conference was consented to participate, and knowledge about being recorded may have influenced their behavior. We collected audio only, and interpreted written transcripts, thus potentially missing nonverbal communication and nuanced verbal cues. This study was conducted at a single institution using only professional interpreters, which may limit generalizability. We did not assess interpretation quality, though all conferences used professional interpreters. Our study focused only on care conferences, excluding other times when families express emotion. Additionally, although we combined 2 comprehensive, validated, and practical frameworks for this analysis, we recognize that they do not capture all emotional or empathic expression. Our all-white coding team was limited in diversity. Patients and families, for individual and cultural reasons, may express emotion or respond to empathy in different ways. For this reason, we worked with an advisory team from various disciplines, backgrounds, racial groups, and cultures to develop the codebook, and check coding and analysis.

Future work should look more broadly at clinician empathy expression across an entire encounter and gather family and clinician perspectives on empathy in conjunction with analysis of conversations. Future work should also explicitly assess how empathic expression impacts shared decision-making and other goals of communication between families and clinical teams.

In summary, our study suggests that, in pediatric care conferences, clinicians often miss opportunities to express empathy, or express empathy that is buried by medical talk. This may impact how families respond and how communication progresses. Our results offer guidance for clinical teams seeking to effectively communicate with families across languages: listening for emotions, employing NURSE(S) empathic statements, and stopping to allow for a response may help deepen discussion. We found no differences in empathy expression or caregiver response to empathic statements for conferences in English compared with those with language interpretation. Additional studies are needed to better understand disparities for families who use language interpretation, and our findings should drive future work to further explore other potential differences in communication that may contribute to disparate experiences and clinical outcomes.

NURSE(S)

name, understand, respect, support, explore, silence/continuer