Emotional labour, emotional regulation strategies, and secondary traumatic stress: a cross-sectional study of allied mental health professionals in the UK

ABSTRACT

The present study investigates relations among emotional labour, emotional regulation strategies, and secondary traumatic stress (STS) in a sample of allied mental health professionals (AMHPs) in the UK. It empirically examines the relationship between emotional labour and STS; and, explores the moderational role of emotional regulation strategies. Participants included 99 clinical psychologists, psychotherapists, counsellors, and psychiatric social workers who completed an anonymous online questionnaire consisting of items related to demographics, emotional labour (surface acting and deep acting), emotional regulation strategies (cognitive reappraisal and expressive suppression), and STS. Fifty-one percent of participants reported high levels of STS. Data analysed using hierarchical multiple regression revealed that age, surface acting, and expressive suppression significantly predicted STS. Deep acting predicted STS only for those participants who reported high levels of STS. Cognitive reappraisal and expressive suppression did not moderate the relationship between emotional labour and STS. Limitations of the study mainly relate to its small sample size. It contributes to the literature by highlighting high levels of STS among AMHPs and providing a rationale for future research on the construct. In addition, it promotes the development of AMHPs’ personal capabilities and professional resources to ensure effective delivery of mental health services.

KEYWORDS:

• Emotional labour
• emotional regulation
• secondary traumatic stress
• allied mental health professionals
• mental health services

Introduction

Mental health ailments constitute a serious public health concern in the UK (Baker, 2018; McManus et al., 2016). The Adult Psychiatric Morbidity Survey (McManus et al., 2016) reports that one in six people are victims of common mental disorders and one in three people seek mental health treatment such as psychotropic medications or psychological therapies. The survey also reports that the percentage of people availing mental health services has increased from 24% in 2007 to 39% in 2014 (McManus et al., 2016). To help address clients’ mental health problems and formulate an effective therapeutic alliance, allied mental health professionals (AMHPs) (such as, clinical psychologists, psychotherapists, counsellors, and psychiatric social workers) empathise with clients (Council on Social Work Education, 2015; Rogers, 1967, 1995; Sharf, 2012). They engage with clients’ emotionally laden narratives which can be stressful and anxiety-provoking yet they have to remain calm and empathetic (Kanno & Giddings, 2017; Smith et al., 2007). They supress or alter their internal feelings and/or external expressions to achieve congruence with clients’ feelings and expressions (Delgado et al., 2017). This process is known as emotional labour (Ashforth & Humphrey, 1993; Hochschild, 1983).

Emotional labour is associated with adverse consequences for AMHPs’ mental health and wellbeing which can consequently affect the delivery and outcome of treatment (Dutton & Rubinstein, 1995; Lawson, 2007; Ruiz-Junco, 2017). It poses a potential risk for the development of secondary traumatic stress (STS); a syndrome that mimics the symptoms of post-traumatic stress disorder (PTSD) but unlike PTSD, is caused by secondary exposure to traumatic events (Baird & Kracen, 2006; Figley, 1995; Stamm, 1997). According to Figley (1995), STS involves ‘natural consequent behaviors and emotions resulting from knowing about a traumatizing event experienced by a significant other-the stress resulting from helping or wanting to help a traumatized or suffering person’ (p. 7). Sprang et al. (2019) postulated that emotional regulation could mitigate the advancement of STS in AMHPs forestalling undesirable consequences for AMHPs and clients. This contention is consistent with the job demands-resources (JD-R) model (Bakker & Demerouti, 2007) which suggests that employees’ personal resources or positive psychological traits attenuate the negative impact of job demands on employees’ psychological health and wellbeing (e.g., Moloney et al., 2018).

Exploring the relation between emotional labour (job demand) and STS (psychological consequence) is important for understanding the impact of providing mental health services on AMHPs’ mental health and wellbeing. Further, investigating emotional regulation strategies (personal resources) as moderating factors aids in the determination of how this relationship can be influenced to abate the risk of developing STS.

Emotional labour and secondary traumatic stress

Emotional labour as construed by Hochschild (1983) comprises of two components; surface acting and deep acting. Surface acting involves discrepancy between internal feelings and external expressions with regards to the other person’s emotive state (Hochschild, 1983). Deep acting on the contrary, involves immersion in the other person’s emotions, achieving correspondence between one’s own and the other person’s feelings and expressions (Grandey, 2000; Smith & Cowie, 2010). Both the forms of emotional labour are associated with different outcomes. Whilst surface acting is associated with stress, burnout, and emotional exhaustion (Fisk & Friesen, 2012; Jirek, 2015; Kenworthy et al., 2014; Schmidt & Diestel, 2014), deep acting is associated with job satisfaction, patient satisfaction, and connection with patients (Chou et al., 2012; Golfenshtein & Drach‐Zahavy, 2015). However, these associations have been found among employees belonging to occupations other than allied mental health services, such as medical or paramedical professions, nursing, and legal advocacy inter alia. Their working conditions and professional norms differ from AMHPs in terms of the extent to which they encourage empathy towards the recipients of their services.

The professional norms of mental healthcare require AMHPs to empathise with clients. Therefore, they engage in emotional labour to facilitate the therapeutic process (Bondarenko et al., 2017). However, due to the accrual or depletion of empathic resources, burnout, vicarious traumatisation, and STS are characterized as normative threats for mental health professionals (e.g., Canfield, 2005; Diehm et al., 2018; Ivicic & Motta, 2017; Jachens et al., 2018; MacRitchie & Leibowitz, 2010). The present study focuses on STS. Its symptoms include depression, insomnia, lack of intimacy with family and friends (Canfield, 2005), intrusive images related to clients’ trauma, avoidance, physiological arousal, stressful emotions, and functional impairment (Bride et al., 2004). It can also negatively impact the quality of care provided to clients (Bride et al., 2007; Choi, 2011; Salston & Figley, 2003; Sexton, 1999) resulting in missed or cancelled appointments, reduced supervision, and increased isolation (Dutton & Rubinstein, 1995). A cross-sectional observational study reported that 70% of British psychotherapists were vulnerable to experience chronic levels of STS (Sodeke-Gregson et al., 2013). In spite of its high prevalence (among AMHPs) and deleterious effects, its relation with emotional labour has not been explored in empirical research.

Moderating role of emotional regulation strategies

Practices generally assumed to alleviate the symptoms of secondary trauma include engagement in leisure and/or self-care activities (Bober & Regehr, 2006). These practices centre on the assumption that time-off work i.e. physical and mental detachment from work completes the cycle of emotions from secondary trauma to homeostasis (e.g., Levine, 1997; Verduyn et al., 2011). However, the successful implementation of these practices require a large workforce of mental health professionals which at present does not exist in the UK (Migration Advisory Committee, 2019). In its absence, it is imperative to focus on individual-level variables to guard the mental health of the existing workforce. Empirical studies suggest that individual-level variables such as, resilience (Harker et al., 2016) and engagement in mindfulness practices (e.g., Hevezi, 2016) reduce the risk of developing STS. One such variable that has been postulated to act as a protective factor (for AMHPs) includes emotional regulation (Sprang et al., 2019).

Emotional regulation refers to the ‘process through which a person modulates his/her emotions elicited by daily events, consciously and unconsciously, in order to reduce their intensity and to respond appropriately to different environmental demands’ (Măirean, 2016, p. 962). It is distinct from emotional labour in the sense that it is a relatively stable tendency (John & Gross, 2004; Liu et al., 2010) whereas, emotional labour is a state dictated by organisational display rules (Diefendorff et al., 2011). Gross’ model of emotional regulation includes two emotional regulation strategies; cognitive reappraisal and expressive suppression (Gross, 1998). Cognitive reappraisal involves consciously altering one’s thoughts about a situation to change its meaning and emotional impact (Gross, 1998). Expressive suppression on the contrary, involves active inhibition of overt emotional expressions to reduce distress (Gross, 1998; Gross & Thompson, 2007). Similar to surface acting and deep acting, cognitive reappraisal and expressive suppression are associated with different outcomes. Cognitive reappraisal is associated with enhanced job performance (Lazányi, 2010), reduced psychopathology (Aldao et al., 2010), and improved psychological well-being (Matta et al., 2014). Whereas, expressive suppression is associated with reduced positive affect and satisfaction with life, and high levels of depression and anxiety (Langner et al., 2012; Lemaire et al., 2014).

The relation between emotional regulation strategies and STS in AMHPs is not widely researched in scientific literature. Only one study was found that suggested that difficulties with affect regulation in Australian clinical psychologists were positively associated with symptoms of STS (Finlay-Jones et al., 2015). STS shares symptomatology with primary traumatic stress (American Psychiatric Association, 2013; Bride, 2007; Salston & Figley, 2003). The relation between primary traumatic stress and emotional regulation strategies is well established in research (Bardeen et al., 2013; Bonn-Miller et al., 2011; Vujanovic et al., 2011). In particular, cognitive reappraisal is associated with reduced primary traumatic stress and expressive suppression is associated with higher levels of primary traumatic stress (Boden et al., 2013; Ehring & Quack, 2010; Măirean, 2016; Moore et al., 2008; Turliuc et al., 2015). However, the moderational role of emotional regulation strategies in the relationship between emotional labour and STS has not been explored in empirical research. The objective of this study is to fill this gap in the literature.

Study aims and hypotheses

The objectives of this cross-sectional study are twofold.

• Firstly, to explore the relation between emotional labour (surface acting and deep acting) and STS in a sample of AMHPs based in the UK.

• Secondly, to investigate emotional regulation strategies (cognitive reappraisal and expressive suppression) as moderator variables in the postulated relationship.

Figure 1 provides a visual representation of postulated relations among the independent, moderators, and dependent variables.

Figure 1. Postulated relations among study variables

Hypotheses for the present study are the following:

H1a: Surface acting will be positively associated with STS.

H1b: Cognitive reappraisal moderates the strength of the association between surface acting and STS, such that the relation will be weaker under frequent use of cognitive reappraisal than under less frequent use.

H1c: Expressive suppression moderates the strength of the association between surface acting and STS, such that the relation will be stronger under frequent use of expressive suppression than under less frequent use.

H2a: Deep acting will be positively associated with STS.

H2b: Cognitive reappraisal moderates the strength of the association between deep acting and STS, such that the relation will be weaker under frequent use of cognitive reappraisal than under less frequent use.

H2c: Expressive suppression moderates the strength of the association between deep acting and STS, such that the relation will be stronger under frequent use of expressive suppression than under less frequent use.

Materials and method

Study design and data collection

An online cross-sectional quantitative survey was carried out using Jisc Online Surveys between June and July 2019. An a priori power analysis conducted using G*Power (Faul et al., 2009) indicated that a total sample comprising of 99 participants would be adequate to achieve sufficient power (0.80) for statistical analyses. This was calculated assuming a medium effect size (f² = 0.15) and an alpha level of 0.05. To be included in this study, participants had to be; (a) practitioners or practitioners in training with active caseloads; (b) work directly in the field of mental health services (such as, clinical psychology, counselling, psychotherapy, or psychiatric social work); and, (c) be based in the UK. AMHPs working with non-traumatised populations were also included in the study as past evidence reports prevalence of STS in mental health professionals with specialisations other than trauma therapy or traumatology (Devilly et al., 2009; Van Minnen & Keijsers, 2000).

Following ethical approval by the institutional review board, participants were recruited via a number of approaches. Trainee practitioners enrolled in the professional doctoral programme in clinical psychology (DClinPsy) at a public university in East Midlands, England were approached to participate in the study. Practitioners were recruited via advertisements on Facebook and LinkedIn pages of AMHPs’ professional bodies such as, UK Council for Psychotherapy (UKCP), British Psychoanalytic Society, British Association for Counselling and Psychotherapy (BACP), British Psychological Society (BPS), and British Association of Social Workers (BASW). To maximise the sample size, convenience and snowball sampling techniques were used in conjunction with the above two strategies. All potential participants were provided with the details of the study and a weblink to the online survey. No incentives were offered to those who chose to participate. The total response rate could not be determined due to the electronic advertisement of the survey.

Measures

Demographic information

Surveyed participants provided information about sociodemographic details such as, age, gender, occupational status (practitioner or trainee practitioner), educational qualification, clinical experience (in years), and caseload volume (average number of clients in a week).

Emotional labour

The Emotional Labor Scale (ELS; Brotheridge & Lee, 2003) was used to quantify two aspects of emotional labour; surface acting and deep acting. Six items (three items for each aspect) were rated on a five-point Likert scale (1, ‘never’ to 5, ‘always’) with a response stem, ‘On an average day at work, how frequently do you … ’. Sample items include, ‘resist expressing my true feelings’ (surface acting), ‘make an effort to actually feel the emotions that I need to display to others’ (deep acting). Responses on both the subscales were summed to obtain two composite scores for each participant. Both the subscales demonstrated satisfactory levels of internal consistency: surface acting (α = .86); and, deep acting (α = .86).

Secondary traumatic stress (STS)

Secondary traumatic stress (STS) was measured using the Professional Quality of Life Questionnaire (ProQOL; Stamm, 2010). ProQOL has been used in several studies and demonstrates satisfactory psychometric properties (Avieli et al., 2016; Connally, 2012; De La Rosa et al., 2018; Stamm, 2010). It includes 30 items that measure compassion satisfaction, secondary traumatic stress, and burnout. For the purpose of the present study, only 10 items pertaining to secondary traumatic stress were included. Items were rated on a five-point Likert scale, ranging from 1 (never) to 5 (very often). Responses were summed to obtain a composite score for each participant. Based on the cutoff scores provided by Stamm (2010), participants could be classified into low, moderate, or high reference categories. However, a review of studies (n = 30) by De La Rosa et al. (2018) suggested that the cutoff scores provided by Stamm (2010) were no longer accurate and thus, suggested revised cutoff scores- 13, 17, and 21-for stratifying participants into low, moderate, or high categories. In the present study, the revised cutoff scores were used. Cronbach’s alpha for the scale was satisfactory (α = .84).

Emotional regulation strategies

The Emotion Regulation Questionnaire (ERQ; Gross & John, 2003) was used to measure participants’ tendencies to use cognitive reappraisal (α = .80) and expressive suppression (α = .82). ERQ includes 10 items; 5 items for each emotional regulation strategy. Sample items include ‘When I want to feel less negative emotions (such as sadness or anger), I change what I’m thinking about’ (cognitive reappraisal), ‘I control my emotions by not expressing them’ (expressive suppression) (Gross & John, 2003). Participants rated the frequency of each item on a seven-point Likert scale, ranging from 1 (strongly disagree) to 7 (strongly agree). Responses were summed to obtain two composite scores for each participant reflecting frequency of use of cognitive reappraisal and expressive suppression.

Statistical analyses

Missing values and tests for violation of assumptions

SPSS (V.24) (IBM Corp, 2015) was used for data management and statistical analyses. Missing values were addressed using mean value substitution (Tabachnick & Fidell, 2001). However, where a case was missing a significant number of values on any scale, that case was deleted. To identify extreme scores or outliers, scores on all measures were plotted on a boxplot (Field, 2013). Also, all scores were converted into z-scores and those above or below ±3 were eliminated from the data set (Field, 2013). Assumptions underlying hierarchical multiple regression were assessed before analysing the data (Field, 2013). Bivariate scatterplots were reviewed to check for the assumptions of linearity and homogeneity of variance (Field, 2013). Normality of residuals was assessed by evaluating skewness and kurtosis values, probability-probability (P-P) plot, and the Kolmogorov-Smirnov (K-S) test (Field, 2013). Collinearity diagnoses (involving tolerance and variance-inflation factor (VIF)), and Durbin-Watson test were conducted to check for independence of errors and multicollinearity (Field, 2013). All the diagnostics were within acceptable ranges for hierarchical multiple regression analysis.

Hypotheses testing

Descriptive statistics (mean, range, and standard deviation) were analysed. Subsequently, relations among study variables were examined using bivariate two-tailed parametric correlation. Hierarchical multiple regression was conducted to test hypotheses. In step 1, sociodemographic covariates (age, gender, occupational status, educational qualification, clinical experience, and caseload volume) were entered as control variables. In step 2, components of emotional labour (surface acting and deep acting) were entered. In step 3, emotional regulation strategies (cognitive reappraisal and expressive suppression) were entered. And, in step 4, four interaction terms (surface acting X cognitive reappraisal, surface acting X expressive suppression, deep acting X cognitive reappraisal, and deep acting X expressive suppression) were entered. The predictor variables and interaction terms were standardized to reduce multicollinearity (e.g., Aiken & West, 1991; Frazier et al., 2004) and bootstrapping was performed to produce conservative bias corrected confidence intervals (Field, 2013).

Results

Information provided by four participants was not included in the analyses due to missing data. Approximately 11% of the cases (n = 11) were identified as being outliers. Since their prevalence represented a significant proportion of the sample, the analysis was conducted twice; once with the inclusion and once with the exclusion of outliers. The difference between the results was marginal, therefore the researchers agreed upon including the outliers in the final analysis. In total, data provided by 99 participants were used for statistical analyses. Age of the participants ranged from 22 years to 69 years (mean = 36.25 years, SD = 11.44 years). Of the sample, 76.47% (n = 65) were females, 61.17% (n = 52) were practitioners, and 82.35% (n = 70) held a post-graduate degree. Years of experience in clinical practice averaged 8.20 years (SD = 7.51 years) and the average caseload volume was 12.0 clients a week (SD = 7.0 clients).

Prevalence of STS

Scores on the STS subscale of ProQOl (Stamm, 2010) ranged from 10 to 39, with an average score of 21.74 (SD = 5.45). Thirteen per cent of the sample (n = 13.13%) reported low levels of STS, 35.35% (n = 33) reported average levels of STS, and 51.51% (n = 51) reported high levels of STS. A demographic breakdown of participants with respective STS scores is presented is Table 1.

Table 1. Sample distribution with STS scores

Sample Characteristics	N (%*)	Secondary Traumatic Stress (STS) Scores
		Low (13–16)	Average (17–21)	High (≥21)
Total Sample	99 (100%)	13 (13.13%)	35 (35.35%)	51 (51.51%)
Gender
Male	23 (23.23%)	1 (4.35%)	10 (43.48%)	12 (52.17%)
Female	75 (75.76%)	11 (14.67%)	19 (25.33%)	45 (60%)
Prefer not to say	1 (1.01%)	-	1 (100%)	-
Occupation
Practitioner	60 (60.61%)	8 (13.33%)	16 (26.67%)	36 (60%)
Trainee Practitioner	39 (39.4%)	4 (10.26%)	14 (35.9%)	21 (53.85%)
Educational Qualification
Missing Information	1 (1.01%)	-	-	1 (100%)
Professional Diploma	8 (8.08%)	-	2 (25%)	6 (75%)
Graduate	10 (10.11%)	1 (10%)	4 (40%)	5 (50%)
Post-Graduate	80 (80.81%)	11 (13.75%)	24 (30%)	45 (56.25%)

*After excluding missing values

Descriptive findings

Descriptive statistics for study variables and bivariate two-tailed parametric correlations between predictors and the dependent variable are presented in Table 2. Correlation analyses revealed low to moderate relations among STS, age, surface acting, and expressive suppression. Age was negatively related to STS, r = −.264, 95% BCa CI [−.433, −.066], p = .015. Surface acting was moderately associated with STS, r = .416, 95% BCa CI [.221, .593] p = .000. Expressive suppression shared a weak association with STS, r = .270, 95% BCa CI [.051, .467], p = .012. Deep acting was unexpectedly not significantly related to STS, r = .168, 95% BCa CI [−.045, .366], p = .123. However, when the cases were split based on scores on ProQOL (Stamm, 2010), it was found that deep acting correlated with STS but only for those participants who scored highly on the scale, r = .390, 95% BCa CI [.120, .629], p = .010. For participants with low and/or moderate STS scores, deep acting was not associated with STS.

Table 2. Mean, standard deviation, and inter-correlations among study variables

	Variable	Mean	Range	SD	1	2	3	4	5	6	7	8
1	Age	36.25	47.00	11.44
2	Gender^+	-	-	-	−.16
3	Occupation^+	-	-	-	−.47**	.13
4	Educational Qualification^+	-	-	-	.06	.17	.21
5	Clinical Experience	8.20	38.50	7.51	.65**	−.01	−.48**	.12
6	Caseload Volume	12.15	31.00	6.63	.09	−.01	−.15	.13	.10
7	Surface Acting	8.42	9.00	2.35	−.17	.11	.00	−.05	−.13	.13
8	Deep Acting	9.22	12.00	2.99	−.11	−.16	.02	.04	−.14	−.09	.15
9	STS	21.76	22.00	5.45	−.26*	−.05	−.06	−.14	−.09	−.14	.42**	.17

*p <0.05 level (two-tailed), **p <0.001 level (two-tailed), +Gender, occupation, and educational qualification are categorical variables

Main findings

Regression coefficients for each predictor on each step of the hierarchical multiple regression analysis are presented in Table 3. Results of the hierarchical multiple regression analysis showed that all the regression models were statistically significant (all ps < 0.05). In the first model, sociodemographic covariates were regressed on STS. The model accounted for 14.9% of variance in STS, F (6, 99) = 2.271, p = .045. Post-hoc power analysis revealed that the model had 80.4% power to detect an average effect (Cohen, 1988), f² = .18. Among the sociodemographic covariates, only age was significantly associated with STS, β = −.435, t = −3.032, 95% BCa CI [−3.863, −.779], p = .002.

Table 3. Prediction of AMHPs secondary traumatic stress by demographic variables and components of emotional labour

Predictors	β	T	F	R^2	Δ R^2	p	BCa 95% CI
							LL	UL
Step 1 (control variables)			2.271	.149		.045
Age	−.435	−3.302*				.003	−.335	−.074
Gender	−.084	−.783				.436	−3.406	1.596
Occupation	−.212	−1.620				.109	−5.281	.502
Educational Qualification	−.056	−.499				.619	−2.863	1.689
Clinical Experience	.113	.779				.438	−.133	.247
Caseload Volume	−.134	−1.254				.213	−.315	.082
Step 2			4.285	.311	.162	.000
Surface Acting (SA)	.398	3.986**				.000	.468	1.364
Deep Acting (DA)	.077	.772				.442	−.164	.472
Step 3			3.348	.312	.001	.001
Cognitive Reappraisal (CR)	−.023	−.211				.834	−.369	.274
Expressive Suppression (ES)	−.011	−.089				.930	−.280	.268
Step 4			2.239	.318	.006	.011
SA*CR	.362	.374				.710	−.163	.183
SA*ES	−.411	−.601				.550	−.157	.091
DA*ES	.230	.248				.805	−.084	.114
DA*CR	−.139	−.242				.810	−.111	.109

*p <0.05 level, **p <0.001 level

Relationship between emotional labour and STS

Hypotheses H1a and H2a predicted that surface acting and deep acting would be positively associated with STS. Therefore, in the second model, components of emotional labour (surface acting and deep acting) were regressed on STS. The model explained 31.1% of variance in STS, F (8, 99) = 4.285, p = .000. It detected a large effect with 99.6% power (Cohen, 1988), f² = .45. Surface acting (β = .398, t = 3.986, 95% BCa CI [1.134, 3.209], p = .001) emerged as a significant predictor of STS. Thus, hypothesis H1a was supported. Deep acting on the contrary, was not significantly associated with STS, β = .077, t = .772, 95% BCa CI [−.461, 1.360], p = .382. Thus, hypothesis H2a was not supported. However, after splitting the cases based on STS scores, it was found that for participants with high STS scores, both surface acting (β = .428, t = 2.227, 95% BCa CI [.214, 3.187], p = .033) and deep acting (β = .341, t = 2.236, 95% BCa CI [.210, 3.760], p = .032) were significant predictors.

Relationship between emotional regulation strategies and STS

The third model examined the variance explained by strategies for emotional regulation (cognitive reappraisal and expressive suppression). The results showed that cognitive reappraisal and expressive suppression accounted for 2.18% of additional variance in STS, F (10, 99) = 3.348, p = .001. The observed effect size was large (Cohen, 1988) with 99.3% power, f² = .45. However, both the emotional regulation strategies: cognitive reappraisal (β = −.023, t = −.211, 95% BCa CI [−1.377, 1.060], p = .836); and, expressive suppression (β = −.011, t = −.089, 95% BCa CI [−1.402, 1.363], p = .924) were not significantly associated with STS.

Moderational role of emotional regulation strategies

To examine the moderational role of emotional regulation strategies in the relation between emotional labour and STS, four interaction terms were included in the fourth model. The results indicated that the model accounted for only 0.6% of incremental variance in STS, F (14, 99) = 2.329, p = .011. Post-hoc power analyses indicated that the model detected a large effect (Cohen, 1988) with 98.5% power, f² = .47. Regression coefficients for all interaction terms were non-significant (i.e. all ps > .05). Thus, hypotheses H1b, H1c, H2b, and H2c were not supported.

Discussion

A small but growing body of research has examined relations among emotional labour, emotional regulation strategies, and secondary traumatic stress (STS; Caringi et al., 2012; Colombo et al., 2019; Leinweber & Rowe, 2010). This job demand – personal resource – psychological cost relationship is evident in human service professionals such as nurses, midwives, lawyers, social workers, educational professionals, and medical practitioners. However, it is important to explore its nature and interaction in mental health workers to ensure effective delivery of mental health services (Figley, 1995). In consideration of this, the present study investigated the association between emotional labour (surface acting and deep acting) and STS, and the moderational role of emotional regulation strategies (cognitive reappraisal and expressive suppression) in a sample of allied mental health professionals (AMHPs) practicing in the UK.

Of the sample, 51.51% of participants were categorised as ‘high’ on STS. It is lower than the prevalence reported by a previous study on psychotherapists employed by the National Health Service (NHS), UK (Sodeke-Gregson et al., 2013). The percentage of trainees experiencing chronic levels of STS in the present study was higher in contrast with a previous study on psychotherapists in training (Beaumont et al., 2016). It is imperative to note that these comparison figures are drawn from studies with different sample sizes and participant characteristics, data collection methods, measures or scales used to quantify variables, and cut-off points for classifying participants into different categories. Particularly, the use of different cut-off points is a noteworthy distinction since the original cut-off scores of the Professional Quality of Life Questionnaire (ProQOl; Stamm, 2010) differ significantly from the revised cut-off scores (De La Rosa et al., 2018). Despite such methodological differences, the descriptive findings of this study suggest that the provision of mental health services can take a toll on the psychological health and wellbeing of AMHPs. It implicates that STS is a pertinent occupational health issue that necessitates a targeted workplace intervention.

Among the sociodemographic covariates, the results indicated that there was a small but significant effect of age on STS implying that as age increases, the risk of experiencing secondary trauma decreases. This finding is consistent with previous studies examining factors associated with STS among nurses and mental health professionals (Hensel et al., 2015; Sacco et al., 2015; Wu et al., 2016). It suggests that unlike young professionals, older professionals are able to protect themselves against the emotionally exhaustive nature of mental health services. Years of experience in clinical practice could be contributory factor in this relationship. In the present study, however, years of experience in clinical practice was significantly associated with age but was not significantly associated with STS. This suggests that factors associated with age, outside the sphere of workplace could also influence one’s experience of secondary trauma.

To the researcher’s best knowledge, no previous empirical study has examined the association between emotional labour and STS in AMHPs. In this sense, the present study was exploratory as it found that surface acting was positively associated with STS in mental health workers. This finding is consistent with the job demands-resources (JD-R) model (Bakker & Demerouti, 2007) which states that challenging job demands negatively impact the psychological health of employees. The relation between surface acting and STS suggests that AMHPs alter their overt emotions to communicate empathy towards clients but at the same time respect professional boundaries by not getting carried away with clients’ feelings i.e., resist engagement in deep acting. Surface acting can lead to emotional dissonance which can further lead to chronic stress or burnout which is a recognised risk factor for STS (Andela & Truchot, 2017; Andela et al., 2015; Cieslak et al., 2014; Hinderer et al., 2014; Shannonhouse et al., 2016; Shoji et al., 2015). The relationship between surface acting and STS observed in the present study could have been fully or partially mediated by burnout.

The other form of emotional labour, deep acting, was not significantly associated with STS. Post-hoc power analysis reported sufficient power to detect a large effect. This suggests that a relationship could exist between the two variables but it needs to be re-examined in future research with a larger sample and more power to detect a small effect. The low factorial validity of the measures used in the current study could have contributed to the lack of a significant relationship with STS. Scales or measures with compromised psychometric properties negatively impact the observed power and effect size (Field, 2013). It could also be due to the heterogeneity in the sample in terms of the professional groups the participants belonged to. Although the professional norms of different specialities within mental healthcare are similar, there exist differences in terms of their individual purpose and training (Society of Clinical Psychology, n.d.). This could have resulted in variation in terms of degree of exposure to secondary trauma and empathy towards clients. As a consequence, responses on the scales might have differed resulting in a statistically non-significant relationship between deep acting and STS. However, for AMHPs experiencing chronic levels of secondary trauma, the results indicated that deep acting was a significant predictor. This implies that the emotionally exigent nature of mental healthcare results in a transfer of symptoms from clients to professionals. The professionals who deeply feel their clients’ emotions are particularly vulnerable to experience the aversive symptoms of secondary trauma (Miller & Sprang, 2017).

The surveyed participants’ self-reported use of emotional regulation strategies did not operate as moderators in the relationship between surface acting or deep acting, and STS. This finding was unexpected as previous studies conducted on social workers (Badger et al., 2008; Wagaman et al., 2015) suggest that emotional regulation acts as a protective factor against the development of STS. However, it is important to note that the regression model testing the interaction terms was statistically significant and had high power to detect a large effect. Also, the variance explained by it was slightly less than 1% to 3% of variance usually accounted for by interaction terms in social science research (e.g., Chaplin, 1991). Therefore, to prevent over-generalisation of this finding, it is important to replicate this study with higher power to rule out a small effect of emotional regulation strategies on the relationship between components of emotional labour and STS.

Implications

The present study contributes to the scientific literature on STS among AMHPs. It is consistent with the suggestion made by Molnar et al. (2017) regarding advancing research on secondary traumatic stress. It examined the relation between emotional labour and STS, and the moderational role of emotional regulation strategies, which to the best knowledge of the researcher has not been previously explored in empirical research. In addition, it provides additional empirical evidence in favour of the JD-R model (Bakker & Demerouti, 2007) by suggesting that job demands are negatively associated with employees’ psychological health and wellbeing. From a practical viewpoint, it highlights high level of STS among AMHPs. It is a cause of serious concern because STS affects the provision of mental health services (Lawson, 2007) which further negatively impacts the outcomes of treatment for clients (Corey, 2009; Nathan et al., 2013). Also, the high prevalence of STS among trainee professionals highlights the need for including STS training (e.g., Bercier & Maynard, 2015) in professional educational programmes. The positive relationship between surface acting and STS suggests that empathising with clients negatively impacts the mental health and wellbeing of AMHPs. Its effects can be assuaged by providing enhanced clinical supervision (Pryce et al., 2007; Sommer, 2008) and self-efficacy training (Cieslak et al., 2016). Also, AMHPs should be encouraged to engage in self-care activities as it has been found to mitigate the risk of developing STS (Sodeke-Gregson et al., 2013).

Limitations

The results of this study need to be interpreted within the context of its several limitations. Firstly, the small size of the non-random sample affected the factorial validity of scales impairing adequate measurement of constructs and relations among variables. Chi-square (χ²) statistic, a goodness of fit index is usually sensitive to large sample size (n > 350) (Hair et al., 2006). Thus, the small sample size of this study affected the factorial structure of the measures used which could have potentially affected the findings of the study. Secondly, the recruitment of participants using convenience and snowball sampling techniques introduced selection bias in the sample (Neuman, 2014). Thirdly, the electronic advertisement of the study inhibited the determination of response rate limiting the external validity of the findings (Neuman, 2014). Fourthly, the exclusive use of self-report measures made the findings vulnerable to common-method bias however, the researcher controlled for it by producing bias-corrected confidence intervals (Field, 2013). Fifthly, the therapeutic orientation of AMHPs and past history of trauma were not controlled for in the current study. These variables can affect the vulnerability to develop symptoms of STS (Sodeke-Gregson et al., 2013). Therefore, by not including them as control variables, the variance in STS explained by the regression models was compromised. And, sixthly, the study did not control for participants’ professions (viz. clinical psychologist, psychiatric social worker, counsellor, or psychotherapist). Although no previous study has examined inter-occupational differences in experiences of secondary traumatisation among allied mental healthcare providers, their slightly distinct professional roles and responsibilities could have accounted for additional variance in STS. Future research should control for the above mentioned sociodemographic variables while examining the relation between emotional labour and STS.