Coding the negative emotions of family members and patients among the high‐risk preoperative conversations with the Chinese version of VR‐CoDES

Abstract Background Little is known about family members' and patients' expression of negative emotions among high‐risk preoperative conversations. Objectives This study aimed to identify the occurrence and patterns of the negative emotions of family members and patients in preoperative conversations, to investigate the conversation themes and to explore the correlation between the negative emotions and the conversation themes. Methods A retrospective study was conducted using the Chinese version of Verona Coding Definitions of Emotional Sequences (VR‐CoDES‐C) to code 297 conversations on high‐risk procedures. Inductive content analysis was used to analyse the topics in which negative emotions nested. The χ 2 Test was used to test the association between the cues and the conversation themes. Results The occurrence rate of family members' and patients' negative emotions was very high (85.9%), much higher when compared to most conversations under other medical settings. The negative emotions were mainly expressed by cues (96.4%), and cue‐b (67.4%) was the most frequent category. Cues and concerns were mostly elicited by family members and patients (71.6%). Negative emotions were observed among seven themes, in which ‘Psychological stress relating to illness severity, family's care and financial burden’ (30.3%) ranked the top. Cue‐b, cue‐c and cue‐d had a significant correlation (p < .001) with certain themes. Conclusions Family members and patients conveyed significantly more negative emotions in the high‐risk preoperative conversations than in other medical communications. Certain categories of cues were induced by specific emotional conversation contents. Patient Contribution Family members and patients contributed to data.


| INTRODUCTION
Misery, disgust, anger, anxiety, fear, sadness and guilt are identified as basic negative emotions. 1 Negative emotions are common in patients 2,3 and family members when providing care for the patients. 4, 5 Bo et al. 6 found that anxiety and depressive emotions had a detection rate of 11.00% and 28.67%, respectively, in hospitalized patients. A meta-analysis showed that a high prevalence of psychological distress was found among family members 7 , and different studies indicated that patients and their family members present negative feelings verbally or nonverbally during doctorpatient communication. 8,9 However, negative emotions are not easy to identify. 10 The Verona Coding Definitions of Emotional Sequences (VR-CoDES), a coding system that was developed to analyse emotional communication in provider-patient encounters, 11 defines concerns as an explicit and clear verbalization of an unpleasant emotional state and cues as a verbal or nonverbal hint to an underlying unpleasant emotion. 12 Patients' hopes, uncertainties, feelings and concerns are expressed indirectly in various medical interviews. 10 Expressing emotional cues and concerns has certain medical implications, such as allowing patients to put their emotions into words and thereby equipping the doctors to recognize patients' psychological distress. 13 However, physicians tend to miss most cues and concerns along with adopting behaviours that discourage any disclosure. 14 While patients are implicit in their expression of concerns, 15 doctors' interactions with patients are mainly focused on obtaining information useful for diagnosis and treatment, which results in limited exploration of worries or issues relating to the condition of the patients. 16,17 Additionally, family members might be unable to fully process the information if their emotional distress has not been addressed. 18 As negative emotions are correlated with insomnia, 19 poor interpersonal relationship, 20 poor judgement 21 and extreme behaviours, 22 we believe that it is very important to identify negative emotions and explore an appropriate coping strategy in the context of poor doctor-patient relationships in China. Patients and their family members, who have inadequate medical knowledge, might experience more negative emotions when confronting decisions on critical health issues such as cancer diagnosis, treatment options and major operations with high risks. According to a consensus among Chinese hospitals, we define high-risk procedures as operations, treatments or examinations with high possibility of unexpected negative outcomes during or after the procedures, which are often linked to high medical costs and may result in serious complications, unsatisfactory therapeutic effects and higher death rates. In such circumstances, patients and their family members are stressed and worried due to the pressure of huge uncertainty and critical condition of the patient. Thus, issues raised in high-risk preoperative conversations and the effectiveness of the communication are important. The preoperative conversation is a means to provide medical information and also acts as a link to strengthen the connection between doctors and patients. However, addressing the complexity of emotional communication is not an easy task. 11 Reliable and validated scales or coding systems are helpful to identify the emotional expressions and evaluate the responding patterns in the conversation. The VR-CoDES, a coding system that was developed to analyse these kinds of emotional expressions, has been widely applied in many medical settings that involve oncology, general internal medicine, dentistry, paediatrics, psychiatry and general practice. 8,11,13,23,24 Nevertheless, little is known about the negative emotions in preoperative conversations regarding high-risk procedures, and how emotional cues are related to the conversation themes has rarely been explored.
To understand the negative emotional patterns under the highrisk preoperative conversation setting and to fill the evidence gap, the present study aimed to (1)

| Procedure
The high-risk preoperative conversations occurred at the meeting room of the Department of Medical Service in the Third Xiangya Hospital of Central South University between January 2017 and April 2019. A digital audio-recorder was applied to audiotape the conversations, and participants' approvals of recording were obtained at the start. We randomly collected 300 audio records from the database. Of these 300 audiotapes, 3 were excluded due to poor recording quality, and 297 conversations were included and analysed in our study. Authors had access to information that could identify individual participants during or after data collection. Two coders, Meng Yin (M. Y.) and Ya Zhao (Y. Z.), who had been trained and qualified to use the coding manual of the Chinese version of VR-CoDES (VR-CoDES-C), coded the conversations. One conversation was divided into several 'sequences' or 'turns' by the coders, and then they coded the sequences that started with the first mention of emotional cues or concerns and ended coding when the topic shifted. 12 The coding procedure followed four steps.
Step 1, 12 conversations were selected randomly and coded independently by two coders, and then disagreements were discussed to reach consensus.

| Analysis
The interaction analysis was conducted using VR-CoDES-C to identify patients' emotional expressions of cues and concerns. Based on the original VR-CoDES, the Chinese version closely follows the definition of cue and concern, as well as the category division. Cues were divided into seven subcategories ( Table 1). The intraclass correlation coefficient (ICC) was recommended as a reliable measure to examine the inter/intra-rater reliability according to previous empirical research when coding doctor-patient interactions using similar schemes to VR-CoDES. 26 The reliability of the VR-CoDES-C (ICC = .79) was acceptable, and its validity (specificity = 0.99, sensitivity = 0.96) was good. 25 The inductive content analysis 27 was performed to explore the emotional contents of identified cues/ concerns. This analysis included three main phases: preparation, organizing and reporting. In the preparation phase, researchers input the recognized cues/concerns into N-Vivo and made a preliminary interpretation of these emotional contents. In the organizing phase, researchers refined these preliminary interpretations which containing cues/concerns by using descriptive words or a shorter phrase. In this phase, these descriptive words were grouped under higher-order headings. In the reporting phase, these condensed descriptors were then grouped into the overarching final themes. Qualitative data were analysed by N-Vivo 12.0. Family member: It is only an ordinary operation to doctors, but it is an earth-shaking event for us.

T A B L E 1 Definitions of subcategories of cues and concerns with examples in this study
Cue-c Words or phrases that emphasize (verbally or nonverbally) physiological or cognitive correlates (regarding sleep, appetite, physical energy, for example) of unpleasant emotional states.
Family member: I could not sleep in the past few days for thinking about his illness.
Cue-d Neutral expressions that mention issues of potential emotional importance that stand out from the narrative background and refer to stressful life events and conditions Patient: The distance between home and hospital is too far.
Cue-e A patient-elicited repetition of a previous neutral expression (repetitions, reverberations or echo of a neutral expression within a same turn are not included) Patient: I think it's safer to be hospitalized.

Cue-f
Nonverbal expressions of emotion Patient: (sigh) OK, fine… Cue-g Clear expression of an unpleasant emotion, which occurred in the past (more than 1 month ago) or is without a time frame Family member: We all hope that he will get better, but he said he wanted to give up his life for some time.
The statistical analysis was conducted and descriptive statistics were reported using SPSS 24.0 (IBM Corp.). χ 2 Was used to test a possible association between cue-b, cue-c, cue-d and the identified themes, and Cramer's V was used to indicate the degree of correlation. p Values less than .05 were considered statistically significant.

| Ethical consideration
Written informed consents were obtained at the start of the  Table 2.

| Inter-rater reliability and intra-rater reliability
The ICCs for inter-rater reliability were .87, .76 and .71 on the identification of cues/concerns, subcategories of cues and the initiation of cues and concerns, respectively. The ICCs for intrarater reliability were .88, .83 and .74 on the identification of cues/ concerns, subcategories of cues and the initiation of cues and concerns, respectively.

| The duration and frequency of expressions
A total of 255 (85.9%) out of 297 recordings were coded with at least one cue or concern. The average conversation length was 27.09 min, ranging from 7.53 min to 1.17 h. A total of 1483 cues/concerns were identified, with a mean of 4.99 cues or concerns per conversation. No cue or concern was identified for the remaining 42 conversations.

| Expressions of cues/concerns
Cues and concerns, respectively, accounted for 96.4% (n = 1430) and 3.6% (n = 53) of the total conversations. A total of 1430 cues were coded into seven subcategories. The frequencies of different categories of cues and concerns are shown in Table 3. The most common category of cues was cue-b, which accounted for 67.4% (n = 964).

| Correlation between cues and themes
When talking about different themes, family members and patients tended to use different cues to express negative emotions (p < .001, formed an investigation on these kinds of conversations by using the Chinese version of VR-CoDES to identify the negative emotions. We hypothesized that the occurrence rate of negative emotional disclosure would likely be higher than ordinary medical consultations, and there might be some correlation between the VR-CoDE's coding of negative emotions and the conversation topics. In this study, we found that the occurrence rate of negative emotions was very high (85.9%) during the whole process of conversations, and it was higher than in many previous studies, in which the occurrence rates ranged from 50% to 76%. [28][29][30] The average number of cues or concerns per conversation in this study was also higher than that in other studies, whose mean ranged from 1.6 to 4.0, 28,30,31 except for those talks conducted for children 23 or when conveying bad news to cancer patients. 32 We also found that cue-b, c and d were closely related to certain conversation themes, among which 'Worries relating to illness severity, family's care and financial burden' and 'Worries relating to the risks of operation' were the most common ones. This study provided additional evidence to the application of VR-CoDES, and initially explored negative emotions among preoperative conversations by using this coding tool. In this study, we observed a very high occurrence of negative emotional cues and concerns among the talks. This high frequency of negative emotions can be mainly explained by the psychosocial burdens that the high-risk procedures placed on both family members and patients. The average duration of these conversations was nearly half an hour, which is obviously longer than daily healthcare communication in the Chinese context, which is always less than 15 min, 9 and might also be the potential reason for identification of more cues and concerns as there were more opportunities for family members and patients to express emotions. We also observed that the most frequent category of cues identified in this study was cue-b, the verbal hints to hidden concerns such as emphasizing, unusual words, expressions of uncertainties and hope. Even though we know that this result is similar to that reported by many other studiesh, 9,28,32,34  We observed certain themes among the conversations.
'Psychological stress relating to illness severity, family's care and financial burden of family' (30.3%) was the most common theme that indicated patients'/family members' negative emotions.
Psychosocial burden, such as lacking personnel and time to take care of the patients and the high medical cost, can easily cause stress. Similar to an intensive care unit (ICU) study, 35  These results demonstrate that family members and patients presented neutral expressions when they talked about the prognosis and conveyed their demands for more information. This implies hidden worries related to these issues.
In summary, we found that more negative emotions may be Future studies will be needed on identification of negative emotions and healthcare providers' response to further examine the correlation between these emotions and certain topics, and to explore the adequate response.
There are several limitations in our study. First, as this was a retrospective study based on data from years ago, it was not possible to verify the coding outcomes with family members and patients who could not be contacted. Second, emotional expressions of family members and patients may be different, but our study did not conduct an analysis of the possible differences because we were unable to identify the individuals due to the limited paper records of the participants. Third, we did not analyse the social demographic characteristics of the participants for the same reason mentioned above.

| CONCLUSION
In conclusion, we identified a high occurrence of negative