The supervisory relationship from an attachment perspective: Connections to burnout and sense of coherence in health professionals.

Supervision is an essential part of the training and work environment of health professionals, especially of psychotherapists and clinical/health psychologists. However, although the supervisory process is always a relational one and may therefore be influenced by attachment dynamics, the importance of the supervisory relationship for the professional's performance and the well-being of the supervisee has yet to be fully examined. In this cross-sectional observational study, the Experiences in Close Relationships-Revised (ECR-RD; avoidant and anxious attachment), the Supervisory Relationship Questionnaire (SRQ), the Maslach Burnout Inventory (MBI) and the Sense of Coherence Scale (SOC-13) were analysed for 346 (81.8% female) health professionals. Considering professional experience and number of supervision sessions as control variables, a better supervisory relationship negatively predicted burnout symptoms (β = -.31) but positively predicted sense of coherence (β = .31, both p < .01). The final model, including avoidant and anxious attachment as additional predictors, explained 30% of the variance in burnout symptoms and 41% of the variance in sense of coherence. The results underline the importance of the supervisory relationship for the well-being and the professional performance of health professionals. Interactions between the supervisory relationship and underlying attachment parameters should be further explored in future studies.

The Supervisory Relationship Questionnaire (SRQ) was used here for the first time in a German translation.

Although general attachment dynamics of the supervisee are of relevance, the supervisory relationship cannot solely be explained as an attachment bond.

The quality of the supervisory relationship seems to have strong connections to burnout and sense of coherence, whereas general attachment dynamics have a stronger connection to the sense of coherence.

Interactions between the supervisory relationship and underlying attachment parameters should be further explored in future studies.

INTRODUCTION

Health professionals—in particular those in training (Petrowski, Pokorny, Nowacki, & Buchheim, 2013) but also experienced health professionals (Azar, 2000; Gibbs, 2001)—may exhibit emotional and behavioural challenges due to the often demanding nature of their work. However, there is some indication that proper supervision and support decreases the connection between challenging work environments and related stress and burnout (Crawford, Adedeji, Price, & Rutter, 2010). Just as the therapeutic alliance is a key agent of client change, the supervisory relationship can be considered a key agent of supervisee development (Ladany, Ellis, & Friedlander, 1999). One of the factors contributing to the establishment and maintenance of a good supervisory relationship seems to be supervisee attachment processes (Bernard & Goodyear, 2013; Gunn & Pistole, 2012). Consequently, general attachment patterns in the supervisee may influence the supervisory relationship as well as supervisee well‐being and resilience (Mikulincer & Shaver, 2003, 2013).

Supervision

Supervision can be defined as ‘the formal provision by a senior/qualified health practitioner of an intensive relationship‐based education and training that is case focused and which supports, directs and guides the work of colleagues’ (Milne, 2007, p. 440). Although supervision within the health and social service sectors is traditionally conducted between practitioners from the same profession (Bernard & Goodyear, 2013), recent years have shown an increase in interprofessional supervision (for an overview, see Davys & Beddoe, 2015). As almost all psychotherapists take on the role of supervisor at one point in their career (Rønnestad, Orlinsky, Parks, & Davis, 1997), it can be assumed that interprofessional supervision is often provided by this profession. In line with this, supervisors more often report degrees in psychology than other educational backgrounds (Mena & Bailey, 2007; Townend, 2005).

As supervision is often described as incorporating elements of training and treatment, theoretical models of psychotherapy have been replicated in supervision for many years (Fernández‐Alvarez, 2016). Therein, one key difference between supervision and therapy is the focus of inquiry (Angus & Kagan, 2007): Although the focus in therapy is on the client's story and a related increase in self‐knowledge and skills, the focus in supervision is also on the supervisee's relationship with this client, and the first goal of the supervisee's development is to provide a better outcome for the client (Angus & Kagan, 2007).

Although ‘modern’ models of supervision focus more closely on the needs of the supervisees (Fernández‐Alvarez, 2016), there are numerous supervision models that show not only a lack of consistency but also a very limited empirical basis (Simpson‐Southward, Waller, & Hardy, 2017). Overall, as stated by Fernández‐Alvarez (2016), ‘alliance is as important in supervision as it is in psychotherapy’ (p. 7). Although not all models of supervision have included the supervisory relationship, ‘it is widely acknowledged that the supervisory process is a relational one’ (Palomo, Beinart, & Cooper, 2010, pp. 132–133).

Attachment

As one of the most extensively researched theories of human development, attachment theory ‘is now used to conceptualize and study adult and couple relationships, work relationships and relations between larger social groups and societies’ (Mikulincer & Shaver, 2007, p. 4). Formed by early environments, the attachment system is active throughout the lifespan (Bowlby, 1988), adaptively regulating psychological and physiological development (Belsky, 1997). Ideally, it provides a ‘secure base’ from which an individual feels confident to explore the environment and a ‘safe haven’ to retreat to when distress arises (Bowlby, 1982). In line with Mikulincer and Shaver (2007), two basic attachment dimensions can be differentiated: anxious attachment (AX) and avoidant attachment (AV). Although individuals with a secure attachment cope with stressful experiences either by actively seeking support or by utilizing mental representations of previously received support, hyperactivating strategies (AX) involve demanding care (Mikulincer, Shaver, & Pereg, 2003), worry and rumination (Cassidy, 1994). Conversely, deactivating strategies (AV) involve a strong need for self‐reliance (Shaver & Mikulincer, 2005) and high levels of unconscious distress (Shaver & Mikulincer, 2002).

Attachment has been shown to influence the interactions between health professionals and their clients, either facilitating or obstructing the course of treatment (Frías, Shaver, & Mikulincer, 2014). The ability to form strong working alliances can be connected to the health professionals' general attachment patterns (Eames & Roth, 2000; Petrowski et al., 2013; Rubino, Barker, Roth, & Ferson, 2000; Schauenburg et al., 2010), with insecure attachment impairing empathy towards the client (Rubino et al., 2000) as well as treatment progress (Fuertes, Moore, & Ganley, 2019).

Using attachment theory and a ‘chain of security’ metaphor (Mikulincer & Shaver, 2007), Kurtz (2005) describes the importance of a supervisor providing a secure base and a safe haven to facilitate the supervisee's exploration, learning and growth: as they feel supported and empowered, this may consequently enable the supervisee to reflect on their own attachment orientation and to more effectively offer a secure base to their clients (Mikulincer & Shaver, 2007). In addition, health professionals may need supervision on how to work with their client's insecure attachment patterns (Wiseman & Tishby, 2014).

The supervisory relationship

Among the early definitions of the supervisory relationship, Bordin (1983) described it as collaboration for change involving an emotional bond and a mutual understanding regarding the tasks and goals of supervision, whereas Holloway (1995) highlighted the formal, hierarchical nature of the supervisory relationship. More recently, Beinart (2014), Cliffe, Beinart, and Cooper (2016) and Palomo et al. (2010) developed a data‐driven measure of the supervisory relationship from the perspective of the supervisee: the Supervisory Relationship Questionnaire (SRQ) captures its evaluative and educative elements as well as the possibility of supervision providing a safe base, an important precondition for other aspects of the supervisory relationship that parallels the safe base elements vital for the formation of secure attachment (Palomo et al., 2010).

Driver (2005) argues that ‘attachment dynamics will frequently be activated’ in supervision, ‘given that supervision is always a relationship in which vulnerability plays some part’ (p. 70). The consideration of supervisees' attachment patterns can therefore increase the supervisors' understanding of differences in supervisory relationships, that is, why some ‘promote change while others are fraught with conflict’ (Renfro‐Michel & Sheperis, 2009, p. 141). Although several researchers have argued that attachment theory helps to better understand the interpersonal dynamics in the context of clinical supervision (Dickson, Moberly, Marshall, & Reilly, 2011; Neswald‐McCalip, 2001; Pistole & Watkins, 1995; Watkins, 1995; White & Queener, 2003), original research on the influence of attachment on the supervisory relationship is still relatively rare. However, previous research indicated that supervisees show similar attachment dynamics in their close relationships and in the relationship to their supervisor (Foster, Lichtenberg, & Peyton, 2007).

Although insecure attachment in the supervisee generally has a negative impact on the supervisory relationship (Bennett, BrintzenhofeSzoc, Mohr, & Saks, 2008; Deal, Bennett, Mohr, & Hwang, 2011; McKibben & Webber, 2017), specific effects of AX and AV patterns have yet to be fully examined. Overall, compulsive self‐reliance related to insecure attachment in supervisees appears to negatively impact the supervisory relationship (Dickson et al., 2011). In addition, general attachment anxiety in supervisees appears to be connected to an increased use of cognitive distortions and related difficulties with corrective feedback during supervision (Rogers, Luke, Gilbride, & Goodrich, 2019). Consequently, some components of supervision (e.g., corrective feedback) and supervisee resistance may diminish supervisee satisfaction with effective supervision (Goodyear & Bernard, 1998; O'Donovan & Kavanagh, 2014). Conversely, secure attachment in the supervisee appears to be linked to greater overall development during supervision (Foster et al., 2007) as well as a higher working alliance rapport with their supervisor (Renfro‐Michel & Sheperis, 2009).

Overall, results underline a strong relationship of supervisee attachment with the supervisory relationship as well as the satisfaction with supervision (Bennett et al., 2008; Marmarosh et al., 2013). In line with this, the supervisory relationship has also been shown to be a good predictor of supervisee satisfaction with supervision (Cheon, Blumer, Shih, Murphy, & Sato, 2009; Palomo et al., 2010; Ramos‐Sánchez et al., 2002) and to contribute, for example, to supervisee self‐disclosure (Spence, Fox, Golding, & Daiches, 2014; Wilson, Davies, & Weatherhead, 2016). In detail, the quality of the supervisory relationship seems to influence whether supervisees address self‐criticism and to buffer its harshness (Kannan & Levitt, 2017). In line with this, a recent systematic review on the role of supervision in general practitioner training underlined the importance of bond, trust, agreement and clarity on supervisory goals in the supervisory relationship while also advocating a peer‐like, non‐hierarchical relationship that may mediate the risk of a power imbalance (Jackson, Davison, Adams, Edordu, & Picton, 2019). Furthermore, another meta‐analysis that included several studies on the relevance of supervisee attachment patterns concluded that the supervisees' view of the supervisory relationship was positively related to the relationship with the client (Park, Ha, Lee, Lee, & Lee, 2019).

As a ‘useful framework in counselor education and supervision’ (Trusty, Ng, & Watts, 2005, p. 75), attachment theory can consequently provide valuable information for the support of professional and personal development in supervisees. However, it is as yet unclear to what extent the supervisory relationship can be considered an attachment relationship (Watkins & Riggs, 2012), and attachment‐based models of supervision often focus on supervisees still in training or at the beginning of their career (e.g., Fitch, Pistole, & Gunn, 2010). In line with this, professional experience of the supervisee has also been discussed as an important factor influencing the process of supervision and the supervisory relationship (Rønnestad, Orlinsky, Schröder, Skovholt, & Willutzki, 2019): in detail, the professional competency felt by more experienced health professionals likely diminishes potentially threatening aspects of the supervisory scenario, thereby reducing the time needed to establish a supervisory relationship. Furthermore, the supervisory process may begin more quickly with experienced supervisees as they likely have a better understanding of their developmental needs. Lastly, more professional experience of the supervisee likely influences the hierarchy in the supervisory relationship, making it more peer like (Rønnestad et al., 2019).

Well‐being

The links between attachment, supervision and (professional) well‐being may best be explained in reference to critical incidents: as summarized by Pistole and Fitch (2008), these key experiences in professional development strain the existing coping skills by challenging the self, thereby providing valuable opportunities for growth. As the attachment system is activated through the incident's novelty and the associated stressful emotions, the supervisee may seek support (i.e., a safe haven) by the supervisor. Especially with less experienced supervisees, it falls to the supervisor to identify attachment‐related needs and to use appropriate caregiving behaviour. When security is restored, the supervisee can return to exploratory system behaviours (e.g., learning new techniques). Difficulties in this process may occur when insecure attachment in the supervisee impairs their ability to profit from supervision but also when the supervisor is unable to provide appropriate caregiving (Pistole & Fitch, 2008). In this study, we considered two parameters related to the general amount of well‐being of the supervisee, which have also shown to be connected to underlying attachment patterns: sense of coherence as a resource promoting mental health (Antonovsky, 1979, 1987) and burnout as an indicator for adverse stress reactions (West, 2015).

Sense of coherence

Overall, sense of coherence has been connected to several health variables, such as psychological well‐being (Kinman, 2008; Packard et al., 2012), adaptive coping strategies (Amirkhan & Greaves, 2003) and social support (Nilsson, Holmgren, & Westman, 2002), whereas a low sense of coherence has been connected to a higher risk for burnout (Feldt, 1997; Kalimo, Pahkin, Mutanen, & Toppinen‐Tanner, 2003). At the heart of Antonovsky's (1979, 1987) concept of ‘salutogenesis’, sense of coherence is a dynamic, pervasive and enduring confidence that the internal and external stimuli encountered in life are explicable and predictable (comprehensibility), that they can be handled with available resources (manageability) and that the challenges of handling these stimuli are worth the required engagement and investment (meaningfulness; Antonovsky, 1979, 1987).

To date, Mikulincer and Shaver (2013) argue that attachment theory is ‘the best and most evidence‐based conception of how close relationships build a person's sense of coherence, safety and value’ (p. 22). In this, attachment security may contribute to a resilient sense of life's coherence (Mikulincer & Shaver, 2013), as feelings of closeness and social support have been connected to a heightened sense of life's meaning (e.g., Hicks & King, 2009; Steger, Kashdan, Sullivan, & Lorentz, 2008).

Although it was originally thought to be relative in adults, research suggests that sense of coherence develops over the entire life cycle (Eriksson & Mittelmark, 2017). Furthermore, there is already some indication that the sense of coherence can be improved by certain psychosocial interventions, such as talk‐therapy groups (Langeland et al., 2006) or psychosocial counselling (Koutsoukou‐Argyraki et al., 2018). In line with this, supervision may strengthen the supervisee's resilience, for example, by increasing their sense of coherence (Howard, 2008; Pack, 2009). Interestingly, Rønnestad et al. (2019) ascribe a sense of coherence in particular to senior professionals, along with a ‘genuineness in relating to clients’ (p. 216).

Burnout

In general, burnout can be defined as a process whereby an individual's resources are depleted to such an extent that they are no longer able to maintain an intense and meaningful involvement at work (West, 2015). Therein, burnout is commonly conceptualized through three interconnected dimensions: it places the experience of individual strain (exhaustion) within the work context, involving both the individual's perception of self (inefficacy) and others (cynicism or depersonalization; Maslach & Leiter, 2008).

Although several studies found high rates of burnout in health professionals (e.g., Cimiotti, Aiken, Sloane, & Wu, 2012; Morse, Salyers, Rollins, Monroe‐DeVita, & Pfahler, 2012), a study on data from 22 countries also found strong associations between burnout and lack of supervision in medical trainees specializing on psychiatry (Jovanović et al., 2016). However, psychotherapists with more years of professional experience seem to exhibit fewer symptoms associated with burnout (Hiebler‐Ragger, Gollner, Klampfl, Nausner, & Unterrainer, 2017).

Attachment theory may help predict why some health professionals but not others experience burnout: although secure attachment has consistently been associated with lower levels of burnout and AX with higher levels of burnout, results for AV are mixed (West, 2015). Importantly, the mechanisms linking attachment and burnout may include various mediating factors, for example, resilience (Tosone, Bettmann, Minami, & Jasperson, 2010), social support (Mikulincer & Shaver, 2007) or supervisor support (Schirmer & Lopez, 2001). As there is a strong need for interventions to reduce burnout in health professionals, it may therefore be beneficial to focus on an increased quality of the supervisory relationship, including more safe spaces for supervisees (Johnson, Corker, & O'Connor, 2020).

The present study

In this cross‐sectional observational study, we aimed to explore the role of the supervisory relationship in relation to the well‐being of health professionals, specifically from the perspective of attachment theory. In this, we focused on symptoms of burnout as a negative indicator and on sense of coherence as a positive indicator of well‐being. Although the applied Supervisory Relationship Scale includes parameters of attachment to the supervisor (e.g., supervision providing a safe base), general attachment patterns were also assessed, given their assumed connection to all investigated parameters (as detailed above). Furthermore, although general attachment patterns have been linked to various aspects of mental health (including burnout and sense of coherence), it is as yet unclear to what extent and in what context (e.g., regarding the professional experience of the supervisee) the supervisory relationship can be considered an attachment relationship. Consequently, we hypothesized that the quality of the supervisory relationship reported by the supervisees would predict their levels of burnout and sense of coherence and that the underlying general adult attachment patterns would be an additionally relevant predictor.

METHOD

Participants

The sample consisted of 346 participants. Demographic and professional characteristics are displayed in Table 1. Among those with a completed training as a health professional (92.2%), 41.3% were psychotherapists or clinical/health psychologists, 42.8% were social workers or educators, 14.2% were medical professionals (e.g., nurses and physicians) and 20.5% had completed another training related to this field (e.g., ergotherapy and music therapy); 24.5% had completed training in more than one of the described areas.

TABLE 1

Demographic and professional characteristics of study participants (n = 346)

	n a	% a
Age (years; range: 23–80; M/SD)	45.68	11.46
Sex (female)	283	81.2
Romantic relationship (yes)	274	79.2
Training as a health professional
Completed training (yes)	319	92.2
Ongoing training (yes)	151	56.4
Professional experience (years; range: 0–50; M/SD)	18.04	11.97
Additional activity as a supervisor (yes)	121	35.0
Cooperation with other types of health professionals
Barely or none	32	11.3
Average	155	44.8
Very much	152	44.0
Satisfaction with professional future perspectives
Barely or none	32	9.2
Average	128	37.0
Very much	186	53.8
In supervision during the last 2 years (yes)	330	95.4
Predominant supervision setting
Single	123	35.5
Group (outside an institution)	98	28.3
Team (in an institution)	109	31.5
Sessions with current supervisor (range: 1–600; M/SD)	37.86	57.94
Intention for further sessions with current supervisor (yes)	264	76.3

Unless otherwise specified.

Materials

Demographic and professional characteristics

To assess basic demographic characteristics, general professional characteristics and specific characteristics related to supervision, the authors of this study constructed an extensive questionnaire based on relevant research discussed in Section 1 (details may be obtained from the corresponding author).

Quality of the supervisory relationship

The SRQ (Palomo et al., 2010) assesses the six dimensions of the supervisory relationship—‘Safe base’, ‘Structure’, ‘Commitment’, ‘Reflective education’, ‘Role model’ and ‘Formative feedback’—from the supervisee perspective. The 67 items are rated on a 7‐point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). For this study, the English original version of the scale was translated into German by the authors of this study and back‐translated by a bilingual person outside the research group. The back‐translation was approved by the author of the original version (H. B.). Palomo et al. (2010) found coefficient alpha values (internal consistency) of .98 for the total SRQ score and .87–.97 for the subscales. Furthermore, the total SRQ score was found to be positively correlated with other measures of supervision constructs, such as the Working Alliance Inventory‐Therapist (WAI‐T) and the Revised Relationship Inventory (RI; Palomo et al., 2010). The coefficient alpha for the total SRQ score with the current sample was .98.

Attachment

The Experiences in Close Relationships‐Revised (ECR‐RD; Ehrenthal, Dinger, Lamla, Funken, & Schauenburg, 2009) assesses AX and AV in intimate relationships (18 items per scale). It employs a 7‐point Likert scale (1 = absolutely disagree to 7 = absolutely agree). Ehrenthal et al. (2009) found coefficient alpha values of .91 for AX and .92 for AV. Both scales were also found to correlate positively with mood pathology in healthy adults (Hiebler‐Ragger, Falthansl‐Scheinecker, Birnhuber, Fink, & Unterrainer, 2016) as well as psychiatric patients (Wulf, Apolinârio, & Tress, 2012). Furthermore, several studies have established the construct validity of the ECR‐RD scales in relation to predictions derived from attachment theory (for an overview, see Mikulincer & Shaver, 2003). The coefficient alpha with the current sample was .87 for AX and .90 for AV.

The Maslach Burnout Inventory (MBI; Enzmann & Kleiber, 1989) consists of 22 items related to ‘Personal Accomplishment’ (PA), ‘Emotional Exhaustion’ (EE) and ‘Depersonalization’ (DP). Items are rated on a 7‐point Likert scale ranging from 0 (never) to 6 (every day). A higher total MBI score indicates higher levels of burnout. Importantly, the MBI is only a screening tool and is not sufficient for a diagnosis (Maslach, Jackson, & Leiter, 1996). Enzmann, Schaufeli, and Girault (1995) found coefficient alpha values of .84 for EE, .71 for DP and .81 for PA. Although most studies use all three subscales (for a review, see O'Connor, Muller Neff, & Pitman, 2018), some also rely on a total MBI score (Bedi, Courcy, Paquet, & Harvey, 2013; Mészáros, Ádám, Szabõ, Szigeti, & Urbán, 2014). In health professionals, burnout has been connected to workload and relationships at work, whereas role clarity, professional autonomy, fair treatment and regular supervision seem to be protective (for a review, see O'Connor et al., 2018). The coefficient alpha for the total MBI score with the current sample was .87.

The short form of the Sense of Coherence Scale (SOC‐13; Hannöver et al., 2004) consists of 13 items related to ‘Comprehensibility’, ‘Manageability’ and ‘Meaningfulness’. Items are rated on a 7‐point Likert scale ranging from 0 (never) and 6 (very often). A higher total score indicates a greater sense of coherence. Hannöver et al. (2004) found a coefficient alpha value of .86 for the total score. The total sense of coherence score was also found to correlate positively with different aspects of mental and bodily health (Hannöver et al., 2004; Schumacher, Wilz, Gunzelmann, & Brähler, 2000). Furthermore, a strong sense of coherence has been connected to a reduced risk of psychiatric diagnosis (Kouvonen et al., 2010). The coefficient alpha for the total sense of coherence score with the current sample was .84.

Procedure

Participants were recruited via networks of health professionals, related email distribution and social media. In detail, all current and past members of the different specialist faculties of the Austrian Working Group for Group Therapy and Group Dynamics (e.g., Specialist Faculty of Integrative Gestalt Therapy, Specialist Faculty of Group Psychoanalysis and Specialist Faculty of Psychodrama) were contacted via email and invited to participate in the study. In addition, information about the study and the possibility to participate was distributed by the corresponding author via different social media groups targeting health professionals. Individuals were eligible for participation if they were 18 years or older and were currently enrolled in or had already completed a training as a health professional in Austria.

All questionnaires were completed online on a Lime Survey® platform after informed consent was obtained. The study was carried out in accordance with the Declaration of Helsinki. Ethical approval was granted by the Ethics Committee of the University of Graz, Austria.

Data analysis

Pearson's correlation statistics were used to investigate the relationships between study variables. Hierarchical regression analyses were used to examine the influence of the supervisory relationship and attachment on burnout and sense of coherence. To minimize Type I error, alpha level was set to p < .01.

Statistical assumptions underlying hierarchical regression (i.e., linearity, normality, homoscedasticity and independence of residuals) were tested. Residual scatter plots indicated that residuals were normally distributed and had linearity as well as homoscedasticity. Results of the white test also indicate homoscedasticity for both models (p > .01). Durbin−Watson test was >1.8 for both models, indicating an independence of residuals. As the variance inflation factor (VIF) for each predictor was below 10, multicollinearity was not an issue.

RESULTS

Characteristics of supervision

In Table 1, the characteristics of supervision reported by the study participants are displayed. When participating in this study, most health professionals (95.4%) were in supervision during the previous 2 years. Of these, 41.6% had their last supervision session during the previous month. Although the number of supervision sessions with the current supervisor ranged from 1 to 600, 76.3% of participants intended to have further sessions with this supervisor (see Table 1).

Prediction of burnout and sense of coherence

Before performing hierarchical regression analysis, the correlations between relevant variables were explored (see Table 2). Higher levels of burnout were related to lower levels of supervisory relationship quality (r = −.35), higher levels of AX (r = .43) and AV (r = .30) as well as fewer years professional experience (r = −.27; all p < .01). Conversely, higher levels of sense of coherence were related to higher levels of supervisory relationship quality (r = .34), lower levels of AX (r = −.54) and AV (r = −.39) as well as more years professional experience (r = .28, all p < .01). In addition, higher levels of burnout were related to lower levels of sense of coherence (r = −.63, p < .01), whereas higher levels of supervisory relationship quality were related to more supervision sessions (r = .17, all p < .01).

TABLE 2

Correlations between study variables

	α	M	SD	1.	2.	3.	4.	5.	6.	7.
1. Supervisory relationship	.98	389.48	63.03	—	−.19 **	−.18 **	−.35 **	.34 **	.13 *	.17 **
2. Anxious attachment	.87	2.13	0.87		—	.43 **	.43 **	−.54 **	−.15 **	−.01
3. Avoidant attachment	.90	2.17	0.94			—	.30 **	−.39 **	−.05	−.03
4. Burnout	.87	70.06	10.78				—	−.63 **	−.27 **	−.13 *
5. Sense of coherence	.84	24.27	14.15					—	.28 **	.11 *
6. Experience	—	18.04	11.97						—	.13 *
7. Sessions	—	37.86	57.94							—

Note: Supervisory relationship: Supervisory Relationship Questionnaire; anxious attachment and avoidant attachment: Experiences in Close Relationships‐Revised; burnout: amount of burnout related symptoms (Maslach Burnout Inventory); sense of coherence: Sense of Coherence Scale; experience: professional experience (years); sessions: number of sessions with the current supervisor.

p < .01.

p < .05.

For the prediction of burnout and sense of coherence, the extent of professional experience (experience) and the number of sessions with the current supervisor (sessions) were entered as control variables in Step 1. Quality of the supervisory relationship was entered at Step 2 and AX and AV at Step 3 (see Table 3). The hierarchical regression analyses, including all predictors and the control variables, accounted for 30% of the variance in burnout, F (5, 340) = 29.68, p < .01, and 41% of the variance in sense of coherence, F (5, 340) = 47.22, p < .01.

TABLE 3

Hierarchical regression analyses predicting burnout symptoms and sense of coherence

Step and predictor variable	Burnout			Sense of coherence
	R 2	ΔR 2	β	R 2	ΔR 2	β
Step 1	0.09 **	0.09 **		0.09 **	0.09 **
Experience			−.26 **			.27 **
Sessions			−.10			.07
Step 2	0.18 **	0.09 **		0.17 **	0.09 **
Experience			−.23 **			.24 **
Sessions			−.05			.03
Supervisory relationship			−.31 **			.31 **
Step 3	0.30 **	0.13 **		0.41 **	0.24 **
Experience			−.18 **			.18 **
Sessions			−.06			.04
Supervisory relationship			−.24 **			.20 **
Anxious attachment			.30 **			−.40 **
Avoidant attachment			.12 *			−.17 **

Note: Supervisory relationship: Supervisory Relationship Questionnaire; anxious attachment and avoidant attachment: Experiences in Close Relationships‐Revised; burnout: amount of burnout related symptoms (Maslach Burnout Inventory); sense of coherence: Sense of Coherence Scale; experience: professional experience (years); sessions: number of sessions with the current supervisor.

p < .01.

p < .05.

At Step 1, experience predicted burnout (β = −.26) and sense of coherence (β = .27; both p < .01), whereas sessions were unrelated to burnout and sense of coherence throughout all steps. At Step 2, experience was still negatively related to burnout and positively related to sense of coherence. Furthermore, quality of the supervisory relationship was an additional predictor for burnout (β = −.31) and sense of coherence (β = .31; both p < .01). At Step 3, experience and quality of the supervisory relationship were still related to burnout and sense of coherence, although βs were smaller than at Step 2. Furthermore, insecure attachment additionally predicted burnout (AX: β = .30, p < .01) and sense of coherence (AX: β = −.40, p < .01; AV: β = −.17, p < .01). Both models showed an increase in R 2 between Step 2 and Step 3 (burnout: ΔR 2 = 0.13, p < .01; sense of coherence: ΔR 2 = 0.24, p < .01).

As psychotherapists and clinical/health psychologists can be assumed to have a closer relationship with supervision (e.g., already during their professional training) than other health professionals, we also investigated whether this may lead to different predictions of burnout and sense of coherence. However, the results of hierarchical regression analyses including only psychotherapist and clinical/health psychologists—burnout: R 2 = 0.27, F (5, 142) = 9.87, p < .01; sense of coherence: R 2 = 0.36, F (5, 142) = 15.70, p < .01—were highly similar to those including other health professionals—burnout: R 2 = 0.30, F (5, 202) = 16.75, p < .01; sense of coherence: R 2 = 0.41, F (5, 202) = 27.53, p < .01—with the quality of the supervisory relationship and AX being the most important predictors in all models.

DISCUSSION

In this cross‐sectional observational study, we aimed to explore the role of the supervisory relationship in relation to the well‐being of health professionals by applying an attachment perspective. In line with our hypotheses, a better supervisory relationship quality negatively predicted burnout symptoms but positively predicted sense of coherence. Interestingly, the predictive strength of the supervisory relationship quality on burnout symptoms and sense of coherence decreased as attachment parameters were added, with AX having a stronger influence on burnout and sense of coherence than AV.

Attachment and well‐being

In total, our findings underline the relevance of attachment dynamics in the supervisory relationship but also highlight that the supervisory relationship cannot solely be explained as an attachment bond. This supports the notion that the supervisory relationship ‘while having the potential to develop into an attachment bond, might best be viewed as involving an affective component that leads to the evoking of attachment dynamics’ (Watkins & Riggs, 2012, p. 256). However, our results may also indicate that specific relationship to the supervisor is more relevant to the supervisory relationship than the supervisee's general attachment (Bennett et al., 2008). Furthermore, attachment dynamics may be more relevant in supervisees who are still in training. Here, as stated by Bennett and Saks (2006), ‘the circle of security within supervision enables the inexperienced student to develop a professional sense of self and confidence’ (p. 673). With increasing experience and competence, however, the supervisee may no longer profit from an attachment‐based supervision (Pistole & Fitch, 2008). Interestingly, our results also underline that professional experience can predict burnout and sense of coherence independently of the supervisory relationship and supervisee attachment.

In addition, our results underline the value of considering different aspects of well‐being in connection to the supervisory relationship. Regarding resilience in health professionals, sense of coherence may have an advantage over other positive aspects, such as self‐efficacy or hardiness (Schäfer, Becker, King, Horsch, & Michael, 2019) as it uniquely combines important aspects of cognitive, behavioural and motivational resistance (Almedom, 2005; Mittelmark et al., 2017). Importantly, sense of coherence represents a general dispositional orientation that is relatively stable but appears to increase with age (Eriksson & Lindström, 2005), whereas burnout is a concept strongly connected to an individual's professional life (Maslach & Leiter, 2008). These differences may explain our finding that the supervisory relationship has a similarly strong connection to both burnout and sense of coherence but that an individual's general attachment dynamics seem to have a stronger connection to sense of coherence than to burnout. Here, higher levels of AX, involving hyperactivating strategies like demanding care (Mikulincer et al., 2003), worry and rumination (Cassidy, 1994) that amplify distress and fail to regulate emotions (Mikulincer & Shaver, 2007), seem to be especially of importance, whereas AV has a considerably smaller impact.

Future research perspectives

Although the SRQ can be considered as ‘one of the few theoretically sound and psychometrically valid questionnaires for measuring the supervisory relationship’ (Cliffe et al., 2016), its usefulness will have to be further assessed through future research in diverse contexts. For example, although clinical supervision has been widely researched in the training of health professionals (Watkins, 2012), research on the role of supervision in experienced health professionals is relatively sparse (Rønnestad et al., 2019). Related to this, the aim of supervision may develop from providing a ‘holding environment’ for health professionals in training—with the supervisor often in the role of instructor/teacher—to a more peer‐like supervisory relationship for experienced health professionals who aim at ‘maintaining a sense of professional growth and resiliency, while avoiding burnout and stagnation’ (Rønnestad et al., 2019, p. 216). In addition, the possibility of differences in the importance of the supervisory relationship or some of its components for different professions still needs to be explored.

In line with previous research (Beinart & Clohessy, 2009; Cliffe et al., 2016; Palomo et al., 2010), our results suggest that supervisees experience wide variations in the quality of the supervisory relationship (SRQ range: 76–465) that have little connection with the number of supervision sessions. Consequently, a good supervisory relationship cannot be assumed but should be consciously addressed and evaluated in daily practice (Palomo et al., 2010) as it requires the ‘active and intentional participation of both parties’ (Inskipp, 1999, p. 186). Furthermore, a possible influence of informal supervision—i.e., significant conversations about clinical practice with individuals other than an official supervisor (Coren & Farber, 2019; Farber & Hazanov, 2014)—on the supervisory relationship with the official supervisor will also have to be examined. Free of the power differential and evaluative components of formal supervision, informal supervision as more of a ‘holding environment’ seems to be able to provide an important additional resource, especially for trainees (Coren & Farber, 2019). Importantly, supervision always has to be clearly differentiated from a supervisee's personal therapy (Fernández‐Alvarez, 2016).

Although our study underlines the importance of a good supervisory relationship in relation to different aspects of well‐being in health professionals, there is as yet no universal agreement on what constitutes a positive outcome in supervision (Palomo et al., 2010). In this, research that includes other aspects of well‐being may lead to additional insights. For example, some findings suggest that the connection between attachment and burnout is mediated by team cohesion and organizational fairness (Ronen & Mikulincer, 2009) as well as hypersensitivity to social rejection and perceived stress (Ronen & Baldwin, 2010). Existential well‐being may also influence the connection, as it seems to counteract the effect of insecure attachment on mood pathology (Hiebler‐Ragger et al., 2016). As an extension of burnout in health professionals, compassion fatigue may also be of relevance, as it includes the experience of secondary traumatic stress (Adams, Figley, & Boscarino, 2008), whereas ‘work engagement’ may be considered as the opposite of burnout (Maslach & Leiter, 2016) or an independent concept defined by a state of contentment composed of vigour, dedication and absorption (Demerouti, Mostert, & Bakker, 2010). Lastly, the consideration of assessments of workplace well‐being targeting specific groups (e.g., psychological practitioners) may also be useful in understanding the processes of supervision (Summers, Morris, & Bhutani, 2019).

Interestingly, an insecure attachment to the supervisor seems to be related to lower levels of self‐reported but not supervisor‐reported professional development (Foster et al., 2007). Furthermore, dynamics in both directions will have to be taken into consideration in future research, as supervision influences clinical practice, and clinical practice influences supervision: for example, the idea of parallel processes that originated in psychodynamic theory (Bernard & Goodyear, 2013; Morrissey & Tribe, 2001) suggests that the supervisee may unconsciously show similar reactions to the supervisor to the reactions shown by their client (Koltz, Odegard, Feit, Provost, & Smith, 2012). Importantly, there is as yet little empirical support for the effect of supervision on client outcome and therapy adherence (Simpson‐Southward et al., 2017).

Lastly, the influence of different supervision settings (e.g., single, team or group supervision), supervision goals and the professional background of supervisor and supervisee on the quality of the supervisory relationship will also have to be considered in future research. Related to this, a higher frequency and an external setting of supervision, a clear differentiation between different forms of supervision (e.g., clinical vs. line management) as well as a trustworthy relationship with the supervisor seem to be related to higher satisfaction with supervision in psychologists (McMahon & Errity, 2014).

Limitations

Presumably, the most important restrictions regarding our study are the use of self‐report measures, especially regarding attachment (Stein, Jacobs, Ferguson, Allen, & Fonagy, 1998), and the cross‐sectional design targeting different types of health professionals. Consequently, longitudinal research will have to more closely explore the relevance of the supervisory relationship in relation to burnout as well as the possibility of supervision strengthening the supervisee's resilience, for example, by increasing their sense of coherence (Howard, 2008; Pack, 2009). In addition, although the SRQ captures certain aspects of the supervisee's attachment to the supervisor (Palomo et al., 2010), a more thorough assessment of attachment parameters in this relationship may generate further insights into the process of supervision and its outcomes. However, our results also indicate that the assessment of general adult attachment patterns may be of additional relevance, especially in light of the complex connections of general attachment with mental health and social functioning (Mikulincer & Shaver, 2003).

Furthermore, the data from participants who act as supervisors themselves (35%) may have influenced our results. Therein, supervision training might prompt an increased awareness of one's own development as well as of organizational contexts, which in turn may have positive effects on the work with clients and supervisees (van Ooijen & Spencer, 2017). In addition, the contributions of the supervisor to the supervisory relationship as well as the dynamics between supervisor and supervisee need to be further examined. Although supervisor self‐disclosure may have a positive effect on the supervisory relationship and the development of the supervisee (Knox, Burkard, Edwards, Smith, & Schlosser, 2008), little is known about the behaviours and the impact of ‘interpersonally sensitive supervisors’ (Shaffer & Friedlander, 2017). Assessing the supervisor's attachment patterns may also be relevant as—especially with supervisees still in training—the supervisor has greater power and greater experience and therefore could be argued to have a stronger influence on the supervisory relationship (Riggs & Bretz, 2006).

Finally, within‐person variability over time and nonignorable non‐response (e.g., if individuals with high levels of burnout were unwilling to participate) may lead to an over‐representation of healthy individuals in our sample. Further research is therefore warranted.

Conclusions

The present study has implications for supervision research, as well as for supervision training and practice. Our results underline not only that a good supervisory relationship is related to less burnout symptoms and a higher sense of coherence in supervisees but also that the supervisee's general attachment patterns influence these connections. These connections seem to be similarly relevant in psychotherapists and clinical/health psychologists as well as in other health professionals.

In accordance with the original SRQ (Palomo et al., 2010), the German translation used in this study seems to be a valuable instrument to assess the supervisory relationship. Consequently, the supervisory relationship and its influence on the well‐being and professional practice of supervisees should be consciously addressed and evaluated in the daily practice of and research on supervision.

CONFLICT OF INTEREST

The authors report no conflict of interest.