Self-awareness in the training of health care professionals: Validation of a group self-awareness questionnaire among Balint groups in China

Group-related learning processes in relation to oneself and in interaction with others play an important role in the training of health care professionals. However, an instrument to measure self-awareness in professional groups is not yet available in China. The aim of the study is to validate a questionnaire on self-awareness processes among Balint groups in China.


Results
Fully completed questionnaires from n = 650 participants were analysed. Most participants came from Beijing, Guangzhou and Shanghai; most were female; and the average age was 31.3 years old (SD 8.2).
The con rmatory and exploratory factor analysis of the G-SAQ did not show good solutions. Therefore, we developed a model with only one general factor and a Cronbach's alpha = .936. Low-to-moderate correlations between the G-SAQ and the previously validated questionnaires (the GCQ and the GRQ) indicated acceptable convergent and discriminant validity. Most of the responses in terms of selfawareness ranged from "something learned" to "relatively a lot learned". The assessment of the group leaders was correlated with the level of learning effects.

Conclusions
The Chinese version of the G-SAQ has satisfactory reliability and validity among participants of Balint groups. It could be used as a measure of self-awareness processes in the future.

Background
It is now widely accepted that self-awareness is an essential component in the training of mental health professionals, especially psychotherapists [1,2]. The basic idea of self-awareness in the context of psychotherapeutic training is based on the assumption that the participants of the training achieve an awareness of problematic and vulnerable aspects of their own self. The goal is to better cope with these problems and to have a more mature personality. Within the framework of health care professional training, trainees should feel their problems from the perspective of the patient. Self-awareness is therefore an indispensable component of the socialization process of a psychotherapist. However, all health care professionals should have some limited self-awareness about the strengths and weaknesses of their own personalities and about unknown and rejected feelings, thoughts and behaviours when treating patients. A speci c form of self-awareness takes place in a group setting. Compared to individual self-awareness, the following additional learning processes are possible in the group setting through feedback from others, also known as peer support: hearing others' similar experiences; the discovery of one's own blind spots in self-perception; and knowledge about unnoticed projections onto others, also known as transference learning [3].
To date, there have been few studies on group self-awareness in psychotherapeutic training, and there have been no questionnaires that assess the learning processes described above [4]. In China, selfawareness is rarely practiced among groups of mental health specialists and among other clinical staff who are confronted with mental and psychosomatic disorders and problems of patients. However, the Balint group has become very common in China in the last 10 years as a kind of patient-related selfawareness [5].
What is a Balint Group?
Between 1949 and 1954, Michael and Enid Balint developed the concept of the "Balint Group," a training and research method for general practitioners.
In these groups, under the direction of a psychoanalyst, doctors discussed their cases through understanding the transference and countertransference in the doctor-patient relationship. Balint group work enables the presenting doctor to bene t from self-awareness and to learn to bear in mind not only the disease but also the patient's personality and social system. After this work, future encounters between the patient and the doctor are more likely to proceed in a more relaxed atmosphere [6]. The success of a group depends on the group members' willingness to be honest, respectful, and supportive of divergent opinions. The content of the group is con dential. A Balint group may meet for a 3-day Balint workshop or over the course of months or years, and group cohesion and trust develop over time.

Balint groups and self-awareness
The principle of group work with physicians that was developed by Balint has been used for over 50 years to sensitize physicians and other professionals regarding the importance of unconscious aspects in shaping relationships with their patients [7,8]. Despite the different objectives between patient groups working in therapy and Balint groups for health professionals, some processes are expected to similar between the two groups. In particular, the group setting and the resulting dynamics of the individual participants among themselves can be used as a central element of self-awareness and selfdevelopment, as posited in the guidelines of the American Group Psychotherapy Association (AGPA) [9]. The following effective factors from group psychotherapy can be identi ed [10].
Self-disclosure: In the context of a case presentation, very personal concerns and feelings are addressed, to which the group members can react and communicate their view of the patient or the involved doctor.
The group dynamics that are initiated by the patient case allow for a deeper re ection of the feelings and thoughts that arise during the session. From the doctor's perspective, this can lead to an expansion of his or her perception of the patient, his or her view and experience of the patient or even a completely different image of the patient. The doctor learns to understand his or her affective reactions towards certain "patient types", especially countertransference.
Catharsis: Within the framework of the protected group work, the participants can allow themselves to experience the feelings and affect that they have previously fended off. These can be anger or rage towards certain patients or deep sadness and helplessness because the attending doctor can no longer see any possibilities for treatment or healing and thus feels powerlessness.
Feedback: Through the feedback of the group members, blind spots can be identi ed more easily. The physician learns to deal carefully with unloved or hidden sides of his or her own problems. However, the group also offers a new understanding for trying out new behaviours in the protected space of the group, e.g., modi ed Balint work in the form of roleplaying or a sculpture. In this way, the new experience can help doctors behave differently in connection with others.
The personality of the group leader The personality of the group leader is also an important variable for the group process in Balint work. Learning processes within Balint groups are dependent on the experience of the group leader, as is the case in other self-awareness groups [11,12]. The group participants need the group leader to be a therapeutic model. If the group leader is rejected, little bene t is derived from the Balint group. On the other hand, a group leader who is perceived as helpful and competent seems to provide a foundation for dynamic-interactive group processes to take place [13].

Balint work in China
During the past ten years, the quality of the doctor-patient relationship in China has steadily deteriorated. Patients and doctors mistrust each other greatly. Continuing violent attacks by disgruntled patients against doctors and hospital staff are now routine events. Hospitals in China and the medical profession are regarded as life-threatening. Medical students are reluctant to become doctors [14][15][16]. Approximately 500,000 recently trained young doctors have not worked as doctors in the last ve years. The misunderstandings and mistrust between doctors and patients are in uenced by social, cultural and economic factors.
What is especially important for Balint work is that doctor-to-patient communication is poor. Doctors have no speci c training in conversation, and their consultations last three minutes on average for 70-100 patients per day. This results in increasingly frustrating experiences for patients and their families.
Balint group work was already part of the EU project Postgraduate training in psychosocial medicine for medical doctors in China, Vietnam and Laos [17]. Between 2005 and 2008, hundreds of Chinese doctors participated in this training. This training programme largely corresponds to basic psychosomatic care in Germany and includes participation in Balint groups. It soon became apparent that Chinese doctors greatly appreciate Balint group work. During and after the Asia-Link programme, doctors began to conduct their own Balint groups in their hospitals. Balint groups are now widespread across China and are well received.
Research on Balint group work in China is new. There are no validated instruments in Chinese for the investigation of self-awareness processes among Balint groups.
Research questions 1. Is the Group Self-Awareness Questionnaire (G-SAQ) a reliable and valid instrument to measure selfawareness processes in the context of Balint groups in China? 2. Which dimensions of self-awareness are reported by the participants of the Balint groups?
3. Are there associations between the group leadership and learning effects of the participants?

Measures
Following a Balint group session, the participants were asked to ll out the Group Self-Awareness Questionnaire (G-SAQ) [18] and two internationally validated instruments: the Group Climate Questionnaire (GCQ) [19,20] and the Group Relationship Questionnaire (GRQ) [21]. Furthermore, sociodemographic data of the participants and information about their previous experiences with Balint The G-SAQ was used because all other group questionnaires represent patient questionnaires. The questionnaire consists of 14 items plus one additional item: "The group leader was helpful for me". The questions are composed of items from previously validated group questionnaires, namely, the Düsseldorfer Wirkfaktorenbogen [22] (items 5, 6, 7, 10, 12 and 13), the Learning Evaluation Form on group self-awareness [23] (items 2, 3, 4 and 8) and the Gruppenerfahrungsbogen (GEB) [24] (items 1, 9 and 11), as well as one newly constructed item (item 14: "...I have not previously been consciously aware of inappropriate attitudes and convictions and they have now become clear").
The G-SAQ assesses group self-awareness processes in psychotherapeutic training. The participants described the learning effects with regard to perceptions of themselves (N = 6 items) and the new experiences with regard to interactions with others (N = 7 items). The questionnaire includes the following dimensions: -The discovery of the self (4, 6, 9, 11), -Interactions with others (3,5,7,8,12) -Discovering one's own blind spots in self-perception through feedback from others (1,4,5,13) -Insights into unnoticed aspects of one's own social relationship life in the sense of unnoticed projections on others (10,14) -Effects on my professional work (2) On the basis of a sample of 335 group participants (physicians and psychologists in training to become behavioural therapists) from 47 self-awareness groups led by 8 different group leaders, a 3-factor solution that explained 60.3% of the variance was developed by factor analysis (principal component analysis with subsequent varimax rotation) [18]. Three scales were distinguished:

Group Climate Questionnaire
The Group Climate Questionnaire-Short Form (GCQ; [19,20]) is a self-report measure that assesses individual group members' perceptions of the group's emotional climate. It consists of 12 items using a seven-point Likert scale ranging from not at all (0) to extremely (6). The GCQ yields three scales derived via factor analysis : engagement, which indicates a positive working group atmosphere and group cohesion; con ict, which re ects anger and tension in the group; and avoidance, which describes behaviours indicating the avoidance of personal responsibility of group work by members.
According to McClendon and Burlingame [25], the GCQ is the most commonly used measure of group climate, and there are a large number of studies supporting its validity. Although several studies on group processes found the avoidance subscale to be psychometrically problematic, the GCQ remains one of the most commonly used measures in group research [26].

Group Relationship Questionnaire
The Group Relationship Questionnaire (GRQ) [21,27,28] consists of 15 adjective pairs that are used to assess the three dimensions of emotional relatedness (e.g., safe-vulnerable, understoodmisunderstood), activity (e.g., spontaneous-hesitating, thoughtful-lively) and safety (e.g., safevulnerable, comfortable -uncomfortable) on the individual level for each patient. The emotional relatedness subscale represents the level of emotional belongingness to the other group members, the therapist, and the group. Low scores on this scale correspond to a higher degree of isolation. The activity subscale represents the subjective level of engagement and activity of a group member during a session. It has been shown to be associated with a higher participation rate and higher levels of self-disclosure [29]. High values on the safety subscale correspond to a feeling of security among the participants.

Translation procedure
The questionnaires were translated and back-translated into Mandarin Chinese according to a state-ofthe-art translation procedure in accordance with the "ITC-Test Adaptation Guidelines" of the International Test Commission [30]. First, two trilingual (Chinese, English, German) Chinese medical students and one trilingual Chinese physician independently translated the German version of the G-SAQ and the English version of the GRQ and the GCQ into Simpli ed Mandarin Chinese. Through the comparison and discussion of the three Chinese versions of the questionnaire, a nal version of the questionnaire in Chinese was developed.
Thereafter, the Chinese draft of the G-SAQ was blindly back-translated into German by a native German speaker with a Bachelor's degree in Sinology, while the Chinese draft of the GRQ and the GCQ were blind back-translated into English by a native English speaker who is uent in Chinese and holds a master's degree from a Chinese university. These back-translated versions of the questionnaires were reviewed by a bilingual (German and English) professor of psychosomatic medicine in comparison with the original version.
In cases of ambiguity, some words and sentences were modi ed in Chinese by the three original Chinese translators. The second Chinese version was once again back-translated and reviewed until the backtranslation was equivalent in meaning to the original version. Finally, the nal Chinese version was used for this research.

Statistic procedures
Using IBM SPSS (26.0) and MPlus 8.4 software, the data were analysed with con rmatory and exploratory factor analyses (CFA). Because of non-normally distributed items, the analyses were conducted with the robust mean and variance adjusted weighted least squares (WLSMV) estimator. The variables of all questionnaires were assumed to be categorical and ordered. Categorical variables are presented as absolute and relative frequencies. Continuous data are presented as the means and standard deviations. Cronbach's alpha was calculated to assess reliability, and validity was investigated using a stepwise linear regression. A p-value of less than 0.05 (2-tailed) was considered signi cant.

Results
A total of 650 questionnaires were analysed. There were only a few missing responses within the G-SAQ (max. 7, 1.1% in item 15). Therefore, we did not use an imputation procedure within the data analysis. Two missing items within the G-SAQ were allowed when calculating a scale value.
Most participants came from Beijing and Guangzhou; most were female; and the average age was 30.8 years old (SD 8.2, N = 577). A total of 56.1% were doctors, 22.6% were medical students, 12.4% were nurses, 7.1% were psychologists, 0.2% were social workers, and 3.8% had other occupations (multiple selection was possible). They had participated in Balint groups an average of 3.7 times (SD 9.85). Sixty percent of participants participated in a Balint group for the rst time at a Balint conference. Most participants completed the questionnaire on the second day of the Balint conference. At that point, they had already participated in two small group sessions and one large group session.
Results of the con rmatory factor analysis and the exploratory factor analysis of the G-SAQ We tested the given structure of the G-SAQ with a con rmatory factor analysis. The t indices were as follows: Chi-Square Test of Model Fit = 336.467 (df = 41. p < .001); RMSEA (root mean square error of approximation) = 0.105; CFI = 0.936; TLI = 0.914; SRMR (Standardized Root Mean Square Residual) = 0.041. This shows an acceptable but not a truly good t. The biggest problem with this result is the very high correlations between the factors. Scale 1 correlates with scale 2 about r = 0.964 (p < .001), while the correction of scale 1 with scale 3 is about r = 0.920 (p < .001). The correlation between scale 2 and scale 3 with r = 0.817 (p < .001) seems to be "low" but is still too high for different constructs.
Therefore, we performed an exploratory factor analysis with all variables of the questionnaire (without questions 2 and 15). The eigenvalue for the rst (general) factor was 7.545, and that for the second factor was 0.976. A two-factor model would still have highly correlated factors (r = 0.761, p < .001) and some double loadings. Therefore, we de ned a model with only one general factor for this questionnaire by using all items except items 2 and 15. The resulting new scale has a mean of 3.40 (SD = 0.93) with Cronbach's alpha = .940.

Validity
Previously validated scales were used to test the construct validity of the Chinese version of the G-SAQ. The G-SAQ general factor shows small-to-moderate correlations with the engagement (r = .421) and avoiding subscales (r = .321) of the GCQ and with the emotional relatedness (r = .242) and activity subscales (r = .104) of the GRQ. No signi cant correlations were found for the "safety" subscale of the GRQ and the con ict subscale of the GCQ (see Table 1).
Insert table 1 here.
The engagement subscale of the GCQ describes constructive therapeutic work, including a positive working atmosphere (item 1), cognitive understanding (item 2), group cohesion (item 4), and self-disclosure (item 11). Among all six subscales, the engagement subscale shows the highest correlation with the G-SAQ general factor and con rms the overall positive learning effect.
The emotional relatedness and activity subscales of the GRQ represent the level of emotional belongingness to the other group members, the therapist, and the group and the level of engagement and activity of a group member during a session. These subscales also show a positive correlation with the G-SAQ general factor but not as high as the correlation between the engagement subscale of the GCQ and the G-SAQ general factor (see Table 1). In general, the results supported the convergent validity of the G-SAQ.
In a linear regression with the general factor as the dependent variable, only a moderate amount of variance is explained, i.e., 24.7% (R 2 = .0247) (see Table 2). This indicates that the G-SAQ measures a different construct than the other two questionnaires.

Results of the individual items of the G-SAQ
Most responses ranged from three (something learned) to four (relatively a lot learned). Items 3 ("... I am now more aware of how others are doing"), item 1 ("... I have learned something for myself from other group members") and the assessment about the helpfulness of the group leader received the highest ratings (see Table 3).

Insert table 3 here
Associations between the group leadership and learning effects of the participants Item 15 ("The group leader was helpful for me") of the G-SAQ was strongly correlated with the G-SAQ general factor (r = .700, p < .001, N = 643).

Discussion
Reliability A general factor model proved to be the best solution for the G-SAQ. From the perspective of factor analysis, the results of this study support the conclusion that the subscales are highly correlated with each other and di cult to separate in our sample. The satisfactory t of the general factor model justi ed the use of a general G-SAQ sum score for further analyses. The reliability was good (Cronbach's alpha = .940).

Validity
The G-SAQ was expected to be moderately associated with emotional relatedness, activity and engagement, which was con rmed for emotional relatedness and engagement, but not for the activity subscale. The weak correlation with activity as well as the higher correlation with the avoidance subscale of the GCQ can be explained by the large number of inexperienced participants. These "members avoided looking at important issues going on between themselves" (Item 3), "depended on the group leader for direction" (Item 7) and "appeared to do things the way they thought would be acceptable to the group" (Item 9). Higher values on the activity subscale represent high levels of participation and a high level of self-disclosure, what would be unexpected ndings.
The low correlations with other scales and the low explanation of variance in regression analysis indicate that the constructs measured by the G-SAQ are different from the constructs measured by the GCQ and the GRQ.

Results of the self-awareness processes
Although most of the participants in the study were beginners, the learning successes are pronounced in terms of self-perception and feedback from others. The good results may also be explained by social desirability. This explains the differences between the items and may explain the high correlations between the subscales in factor analysis.
A glance at the literature shows that similar learning outcomes are described in earlier quantitative and qualitative studies of Balint groups. Some quantitative studies found an increase in psychosocial selfe cacy, but some studies found these ndings only after long-term participation [31][32][33]. However, the sample sizes were small and not representative.
A large quantitative German study with 1635 participants investigated self-awareness processes in a sample of medical doctors in Balint groups. The results highlighted that Balint group work is highly associated with personal gains in self-awareness and learning processes related to transference and feedback in group work. Notably, one-third of all participants has more than two years of experience with Balint group work [34].
Some qualitative studies describe the following self-awareness processes: a general maturation in participants' defence mechanisms [35,36], "knowledge of one's own limits" and "minimal interference of one's own psychopathology" [37], awareness of one's own and patients' feelings [38], interacting with patients differently, meeting other perspectives [39], and individual change in some participants' approach towards the group and the patients [35]. One limitation is that these are interview data from a small sample of participants (N = 3 to 16).

Group leadership and learning effects of the participants
The importance of group leadership for the learning processes with regard to self-perception and the perception of others is indicated by the high positive correlation between the question of how helpful the group leadership was and the individual items of the questionnaire. In a large quantitative study of the German Balint Society, group leadership had an impact on the learning effects of participants of Balint groups [40]. Effective learning processes in Balint groups are strongly related not only to the method itself but also to the personality of the group leader. Standards for the leadership of Balint groups are available [41,42]. Guidance for the training of Balint group leaders includes creating a safe environment; protecting the presenting colleague from questioning; encouraging the open expression of thoughts, fantasies and physical reactions; avoiding premature and hasty conclusions; enduring silence and uncertainty; and supporting a new way of looking at the doctor-patient relationship.

Limitations
The results are representative of participants after 2-3 Balint group sessions. No statements can be made about the self-awareness processes in a continuous Balint group across months. It is unclear whether these results are generalizable to self-awareness processes in other Balint groups, at other institutions and with a different composition of participants. However, it can already be assumed that the results are not speci c to the sample that has now been investigated. The selected items are all based on previously validated group questionnaires, which have been used in different therapeutic and selfawareness groups with different theoretical-conceptual approaches. The results are also consistent with the existing literature.
The questionnaire was developed for participants of a self-awareness group in psychotherapeutic training. For the further development of the questionnaire, it would be useful to focus more clearly on patient-centred self-awareness. For this purpose, the items with "others" would have to be replaced by "patients". This also applies to the use in other settings with a different background of healthcare professionals.

Conclusions
The Chinese version of the G-SAQ has satisfactory reliability and validity among participants of Balint groups. It could be used as a tool to measure self-awareness processes in the future. The questionnaire warrants further testing in both research and clinical situations. Future research should examine the associations between self-awareness and the quality of the doctor-patient relationship, the development of learning success in terms of self-awareness through continuous participation in a Balint group over months and years, and the associations between learning effects on self-awareness and on the competence of the group leader.

Declarations
Ethics approval and consent to participate