Does attending to extremely poor clients increase the burnout of social workers?

ABSTRACT

This study examines how working with extremely poor clients contributes to social worker burnout. Previous research has shown that the emotional labour social workers perform on a daily basis may have negative consequences such as job stress and burnout. Studies have also reported that clients’ poverty may increase these consequences through several mechanisms. Based on these findings, we hypothesised that social workers working with extremely poor clients will be more likely to burn out than those working with less poor clients. Data from a survey of Hungarian social workers in child and family welfare services were analysed using linear regression models. Burnout was measured using the Maslach Burnout Inventory (MBI) distinguishing three dimensions of burnout: emotional exhaustion, depersonalisation, and personal accomplishment. Results show that working with extremely poor clients is not associated with higher emotional exhaustion and higher depersonalisation, but it significantly decreases the perceived level of personal accomplishment. This finding highlights the importance of addressing the significant chronic stress experienced by those working with the extremely poor, which exposes them to various mental and physical illnesses in the long term. The study also discusses the methodological implications of our findings on the applicability of the MBI.

ABSZTRAKT

A tanulmány azt vizsgálja, hogy a mélyszegénységben élő kliensekkel végzett munka hogyan járul hozzá a szociális munkások kiégéséhez Magyarországon. Korábbi kutatások kimutatták, hogy a szociális munkások által naponta végzett érzelmi munka olyan negatív következményekkel járhat, mint a munkahelyi stressz és a kiégés. Néhány tanulmány arról is beszámolt, hogy a kliensek szegénysége több mechanizmuson keresztül is fokozhatja ezeket a következményeket. Ezen eredmények alapján azt a hipotézist fogalmaztuk meg, hogy a mélyszegénységben élőkkel foglalkozó szociális munkások nagyobb valószínűséggel fognak kiégni, mint a kevésbé szegény ügyfeleket kiszolgáló társaik. A család- és gyermekjóléti szolgálatoknál dolgozó szociális munkások körében végzett felmérés adatait többváltozós regressziós modellek segítségével elemeztük. A kiégést a Maslach Kiégés Leltár segítségével mértük, amely a kiégés három dimenzióját különbözteti meg: érzelmi kimerülés, deperszonalizáció és személyes teljesítmény. Eredményeink alapján a mélyszegény kliensekkel végzett munka nem jár magasabb érzelmi kimerüléssel és deperszonalizációval, mint a kevésbé szegény kliensek kiszolgálása. A mélyszegénységnek való kitettség ugyanakkor növeli a kiégést a szociális munkások körében a személyes teljesítmény alacsonyabb szintjén keresztül. Ez a megállapítás rávilágít annak fontosságára, hogy foglalkozzunk azzal a jelentős krónikus stresszel, amelyet a mélyszegénységben élőkkel dolgozó szociális munkások tapasztalnak, és amely hosszú távon különböző mentális és fizikai betegségeknek teszi ki őket.

KEYWORDS:

• Emotional labour
• burnout
• social workers
• poverty
• Hungary

KULCSSZAVAK:

• érzelmi munka
• Kiégés
• Szociális munkások
• magyarország

Introduction

People living in poverty make up a major segment of child and family social workers' clients. Although poverty may not be their manifest problem, it is often the reason why they turn to or are referred to social services (Waldegrave, 2005). Social workers help impoverished families to make use of social benefits and services (Carlson, 2017; Schiettecat et al., 2017). Beyond material support, they also provide them with emotional support (Featherstone et al., 2014; Lavee & Strier, 2018).

There has been a considerable amount of research on social workers’ perceptions of poverty (e.g. Blomberg et al., 2013; Bullock, 2004; Weiss-Gal et al., 2009) and how these perceptions affect service policy (e.g. Monnickendam et al., 2010; Reingold & Liu, 2008). However, relatively less attention has been paid in the literature to social workers’ emotions when attending to families living in poverty (Lavee & Strier, 2018).

As frontline service workers, social workers are expected to manage their emotions, to fake or suppress an emotional display (e.g. anger, frustration). The emotional labour (Hochschild, 1983) they perform on a daily basis may have negative consequences such as job dissatisfaction, job stress and burnout (Hochschield, 1983; Jeung et al., 2018).

Qualitative research has shown that working with extremely poor clients is particularly emotionally demanding for social workers (Ruiz-Fernández et al., 2021). The aim of this research is to examine whether working with clients in extreme poverty increases social worker burnout in the Hungarian context. Based on the research findings of Lavee and Strier (2018), and Ruiz-Fernández et al. (2021), our hypothesis is that social workers working with clients in extreme poverty will experience more negative consequences of emotional labour and thus will be will be more likely to burn out than social workers working with less poor families.

We analyse the association between clients’ extreme poverty and burnout of social workers working in child and family welfare services in Hungary. The country provides an appropriate context for the analysis of this research question, as – despite a slight improvement in poverty indicators in recent years – it is characterised by a relatively high material deprivation rate and considerable variation in the incidence of extreme poverty between regions and municipalities. Small, remote settlements in North-Eastern Hungary are characterised by high levels of extreme poverty, while other municipalities, such as Budapest and the larger towns have a much lower incidence of extreme poverty.

Our analysis is based on data from an original survey of social workers working in child and family welfare services in Hungary. The three subscales of burnout, emotional exhaustion, depersonalisation and personal accomplishment are measured using a version of Maslach's Burnout Inventory developed for employees working in institutions providing human services (Maslach et al., 1996). The measurement of extreme poverty among clients is based on extreme poverty as perceived by social workers in their service area. We use factor analysis to examine the factor structure of burnout items in our sample. Multiple linear regression models are used to analyse the association between clients’ extreme poverty and the three subscales of burnout, controlling for sociodemographic, work-related and organisational predictors of burnout.

The structure of the study is the following. In the first section, we shortly review the related previous findings and formulate the hypothesis of our study. In the second section, we outline the Hungarian contexts of extreme poverty and social work in child and family welfare services. The third section describes our data and the measurement of the dependent variable, as well as the main independent variable and controls. The main results of our analyses are presented in the fourth section. The last section concludes and discusses implications of the analyses.

Previous findings and hypothesis

Emotional labour in social work

Only recently have studies been published that apply Hochschild’ concept of emotional labour (1983) to social work practice (e.g. Kanasz & Zielińska, 2018; Winter et al., 2019). According to Hochschild, emotional labour ‘requires one to induce or suppress feeling in order to sustain the outward countenance that produces the proper state of mind of others’ (Hochschild, 1983, p. 7.). To act professionally, workers in frontline service occupations, such as social workers, are required to manage their emotions in exchange for wages. This means that the social worker's expression of emotions needs to correspond to what a client expects as appropriate in a given situation (Kanasz & Zielińska, 2018; Moesby-Jensen & Nielsen, 2015). The findings of a qualitative study, conducted among Polish social workers by Kanasz and Zielińska (2018), suggest that social workers are engaged in emotional labour ‘mostly in situations which require self-control of anger and negative feelings toward clients’ (p. 357). Moesby-Jensen and Nielsen (2015) identified three types of emotional labour in Danish social workers. The first type occurs when the social worker shuts off emotions both during and after the meeting with the client, so as to avoid revealing how she or he was affected by the situation. The emotional labour the social worker performs in this case is referred to as ‘surface acting’ by Hochschild (as opposed to ‘deep acting’ which refers to attempting to change emotions). The second type of emotional labour is when the social worker postpones emotions and processes them at a later time. The third type refers to cases that make such a strong impression on the social worker that it affects how she or he manages and relates to the case (‘when a case gets under your skin’).

Emotional labour in the service sector has been linked to various job-related negative consequences, such as job dissatisfaction, job stress, depersonalisation, and emotional exhaustion (Jeung et al., 2018). Hochschild (1983) and other researchers (e.g. Zapf, 2002) have reported that emotional labour is stressful and may lead to burnout. Research has revealed different mechanisms through which emotional labour may contribute to burnout. One of these mechanisms brings surface acting into focus. Surface acting is supposed to result in emotional exhaustion due to the effort needed to fake or suppress negative emotions (Brotheridge & Grandey, 2002; Jeung et al., 2018).

In addition, social work is found to be among the professions characterised by a high risk of client violence (Mayhew, 2002). Conflicts with clients are reported by Savaya (2014) as an important daily job stressor (in addition to conflicts with co-workers and supervisor and job-related dilemmas). Client conflicts may include verbal and physical abuse, violation of boundaries, as well as deception and manipulation.

Emotional labour of social workers attending to families in extreme poverty

Only very few studies have addressed the emotions of social workers working with clients in situation of extreme poverty. A recent work by Ruiz-Fernández et al. (2021) aimed to identify social workers’ emotions and experiences while attending to vulnerable social groups, including people in extreme poverty. The study concludes that ‘social workers experience high levels of emotional discomfort when carrying out their work, which is exacerbated when the populations they attend to are particularly vulnerable groups, such as … people with scarce or a lack of economic resources’ (Ruiz-Fernández et al., 2021, p. 11).

Lavee and Strier (2018) examined the emotional labour of social workers working with impoverished families in Israel. They identified the elements behind the emotional flooding that social workers experience in their professional routines. One element, referred to as ‘institutional abuse’, is related to the institutional context within which social workers carry out their work. When families in extreme poverty turn to social services, the problems they face often require more resources than what the social service system can provide. Social workers have to face the ever-increasing client demands on the one hand, and the lack of adequate material resources and emotional guidance on the other hand. Another element is ‘emotional attrition’ which refers to the strong emotional experiences that come with the intensive work with the poor, including stress, frustration, anxiety, fear, and helplessness. The third element of emotional labour, as identified by Lavee and Strier (2018), is ‘personal precariousness'. As a result of their encounters with the poor, and the realisation that poverty ‘could happen to anyone’, social workers may feel anxiety and concern for their own lives and those of their family members (Lavee & Strier, 2018, p. 4.).

Hypothesis

In light of the previous research findings on service sector workers in general and social workers in particular, it can be expected that social workers working with families in extreme poverty are more likely to encounter situations where they need to perform surface acting for some reason; they are more likely to face situations where the client's needs far exceed the resources available to meet them; they are more likely to come into stressful contact with clients than social workers serving less poor clients. These factors, both on their own and in combination, increase the amount of emotional labour that social workers working with clients in extreme poverty have to perform. For this reason, we hypothesize that social workers attending to clients in extreme poverty are more likely to burn out than those attending to less poor clients. Since the relevant literature does not allow us to formulate specific hypotheses on certain aspects of burnout, our study is exploratory in this respect.

Hungarian context

Extreme poverty is a severe form of poverty that is characterised by a high level of material deprivation and vulnerability. The income of extremely poor families falls so far below the poverty line that it is insufficient to meet basic human needs such as food, shelter and clothing. This type of poverty makes individuals and families extremely vulnerable to hunger, disease, and other life-threatening conditions.

The poverty rate (as measured by the AROP rate) in Hungary is consistently below the EU average, at 12.1% in 2022, compared to 16.5% in the EU27 (Eurostat Database, 2022). However, the evolution of the relative poverty gap suggests that the income of the poorest is increasingly decoupled from the rest of society, making it less possible for them to escape poverty. Moreover, poverty is not just income poverty, but is often associated with material deprivation and social exclusion. Despite a significant decrease in recent years, the share of people living in severe material deprivation at the national level remains much higher than the EU27 average and is particularly high at 26.2% in the lowest income quintile (Eurostat Database, 2022). There has been little change in recent years in the persistent marginalisation of the most vulnerable groups, and the recurrence of extreme poverty and exclusion among them continues (Branyiczki & Gábos, 2019).

In Hungary, marginalised groups are concentrated in rural villages and small towns in old industrial areas (Szirmai et al., 2016). The disparities between the capital and disadvantaged rural areas are very large. The persistence of extreme poverty in the least developed areas is a consequence of the accumulation of disadvantages: it is partly related to structural problems in local labour markets and partly to less developed human infrastructure and unequal access to quality social, health and education services (Ferge, 2010).

Local social services have an important role to play in interrupting the recurrence of social marginalisation and mitigating its consequences. Child and family welfare services have the closest links with families living in poverty. The law requires local authorities to operate such services in all municipalities. However, the human and material resources needed to provide quality service are unevenly distributed (Husz et al., 2020), and it is often in the municipalities with the highest poverty levels that there is insufficient capacity to run the service. A survey shows that in the municipalities with the highest poverty rates, the average number of families served by a social worker is 57, compared to 32 in the least poor municipalities. In addition to high caseloads, excessive workloads and the size of administrative tasks, combined with inadequate training and supervision are also pressing problems in day-to-day work (Husz et al., 2020). In many municipalities, the capacity is only sufficient to deal with crisis situations and provide social assistance, but not for prevention and development activities (Kopasz, 2017). The most disadvantaged situation is in small settlements, where family support is often only available for 2-4 hours per week and is provided by a single person (hereinafter referred to as one-person service) (Gál, 2017). In 2022, 30% of family and child welfare services were provided by a single person (HCSO Database, 2022).

Research shows that exposure to extreme poverty also shapes the attitudes of social workers. In settlements where there are more extremely poor families among clients, social workers are more likely to blame the poor for their poverty than in less poor municipalities (Husz et al., 2022). Individual blame can lead to less supportive attitudes towards them and less active involvement in seeking solutions to their problems.

Materials and methods

Description of data

The data source used for the present analysis is a cross-sectional survey of social workers in family and child welfare services in Hungary, conducted between November 2018 and February 2019. The research conducted is compliant with the research ethics principles of the Institutional Ethics Review Board of the HUN-REN Centre for Social Sciences (approval number 1-FOIG/130-31/2022). Participation in the survey was voluntary and respondents’ declarations regarding informed consent have been collected.

The primary aim of the data collection was to assess the working conditions of social workers, and as part of these data on burnout was also collected. The sampling was based on a list of municipalities where family and child welfare services operated and the number of social workers in each municipality. A multistage stratified sampling method was used, with the first stage involving a random sample of counties to ensure the regional representativeness of the data. The municipalities of the selected counties were then stratified according to the number of social workers and a random sample was drawn from each stratum. The aim was to ensure that the sample contained an appropriate proportion of one-person services. Of the 652 municipalities with family and child welfare services, 178 were included in the sample, covering 9 of the 23 districts of Budapest. The proportion of social workers contacted in each municipality varied between 50 and 100%, depending on the number of social workers in the municipality. A computer-assisted personal interview (CAPI) was conducted with the sampled social workers, with a response rate of 65%. After data cleaning, the database contained 600 individuals, 19% of the total population of social workers in family and child welfare services.

The sample of social workers was dominated by women (90% vs 10% men). In terms of age, half of the respondents were under 40, one-third were aged between 40 and 49, and 19% were aged 50 or over. Over one-tenth of social workers lived in Budapest, while 52% lived in other towns and 36% in villages. All respondents had some higher education: nearly two-thirds had a degree in social work or a related field (e.g. sociology or social policy). Social workers in the sample had on average more than nine years of experience in social work.

Measurements

Dependent variables

For the assessment of burnout, we used a version of Maslach's Burnout Inventory (MBI-HSS) developed for employees working in institutions providing human services (Maslach et al., 1996) and used in previous studies among Hungarian social workers (e.g. Győri & Perpék, 2022). The questionnaire contains 22 statements and respondents can indicate on a seven-point Likert scale how often the statements listed describe their work. The questionnaire measures burnout on three subscales and is organised around the Emotional Exhaustion (EE), Depersonalisation (DP), and Personal Accomplishment (PA) scales. Emotional exhaustion is an individual dimension measuring an individual's physical and psychological workload, while depersonalisation is an interpersonal dimension that measures distancing from work, clients and colleagues. Personal accomplishment is a dimension related to self-evaluation measuring the sense of competence at work. In the questionnaire, emotional exhaustion is measured by 9 items, depersonalisation by 5 items and personal accomplishment by 8 items. Higher values on the subscales of emotional exhaustion and depersonalisation indicate potential burnout, while lower values on the subscale of personal accomplishment indicate burnout. Permission has been received from Mind Garden (Mind Garden Inc., Menlo Park, CA) to use the official Hungarian version of MBI-HSS.

Independent variable

To measure extreme poverty among clients, we used a survey question asking respondents whether they agreed that there were many families living in extreme poverty in their area of service provision. The question could be answered on a four-point scale, where 1 means not true at all and 4 means completely true. This variable was included in the models as a dummy variable, where a value of 1 indicates that the respondent thinks the statement is completely true or rather true, while 0 indicates that it is not true at all, or rather not true.

Control variables. In the analysis, we controlled for several sociodemographic and work-related variables which, according to the literature, may be associated with the development of burnout syndrome. Many studies have found an association between sociodemographic variables (e. g. gender, age) and burnout, although findings have been inconsistent (Listopad et al., 2021; Toledano & Ruiz-Olivares, 2023). In our study, we controlled for gender as a dummy variable, age as a three-category variable (under 40, 40–49 years, 50 years and over), and the presence of a social work or related (e.g. sociology or social policy) degree as a dummy variable. Social work experience was included in the model as a continuous variable. Among the work-related factors that may protect against burnout, a systematic review by McFadden et al. (2015) emphasises the role of manageable workloads, supervisory support, as well as a supportive social environment at work. To measure workload, we used a question asking the respondents’ perceptions of whether they can spend enough time with their clients. Supervisory support was measured by a dummy variable indicating whether the respondent had participated in supervision in the last two months. Another dummy variable was included to measure whether the respondent receives professional support at work. In Hungary, social workers working in one-person services are particularly vulnerable to high workloads and low social support at work; therefore, we included a dummy variable to identify these respondents.

Research on burnout in general and among social workers in particular identifies organisational factors such as reward structures (wages, benefits) and opportunities for job and career development as important factors (Listopad et al., 2021; McFadden et al., 2015). The following variables were used to measure these organisational factors (included in the model as discrete variables): satisfaction with the reward received for work – separately measuring satisfaction with material rewards (salary) and moral esteem – and opportunity for professional development. Social work is often made difficult by the insufficient material conditions of work, as well as the frequently changing legal and bureaucratic rules. Material conditions of work were measured by a dummy variable indicating whether the respondent is provided with a mobile phone (for her or his own use) to communicate with clients¹. Another dummy variable was included to measure whether the respondent finds it difficult to keep up with the changing rules and protocols Finally, we controlled for the type of settlement (categorical variable: capital, towns, and villages), which may be related to various aspects of the work environment. Descriptive statistics of the variables used in the analysis is shown in Table A1 of the Appendix.

Methods

Statistical analysis was performed using STATA 16.0 software. In the first part of the analysis, the review of internal consistency of the MBI was carried out by means of confirmatory factor analysis. When assessing the fit, we expected the relative fit index Comparative fit index (CFI) and the Tucker-Lewis index (TLI) providing comparative fit to be above 0.900, and the root-mean-square-error of approximation (RMSEA) and Standardised root mean squared residual (SRMR) indicators used for absolute fit to be below 0.080 (Hu & Bentler, 1999). In addition, the internal reliability of the scales obtained was checked by calculating the Cronbach-α value. In the second part of the analysis, the correlations between burnout rates and extreme poverty among clients were examined using linear regression models, in which the influence of relevant sociodemographic and work-related factors was filtered out. The dependent variables used in the regression analysis are specific dimensions of burnout (subscales of emotional exhaustion, depersonalisation and personal accomplishment). In the initial linear regression models we included the most important sociodemographic and working conditions variables, and in a next step we added the main independent variable measuring the extreme poverty of clients.

Results

Reliability analysis: MBI – HSS’s internal structure

As the first step, a confirmatory factor analysis (CFA) was performed on the sample of Hungarian social workers, following the recommendations of the MBI-HSS validation study conducted among Hungarian doctors (Ádám & Mészáros, 2012). Since the fit indicators of this model proved to be inadequate, an exploratory factor analysis (EFA) (principal component analysis with Varimax rotation) was carried out to identify the best-fitting model for the social worker sample. The Kaiser–Meyer–Olkin indicator showed a satisfactory value for all 22 items (KMO = 0.926), similar to the Bartlett's test (p < 0.001). The exploratory factor analysis also resulted in a three-factor model; the total variance explained by the three factors with more than one eigenvalue is 64.8% (ꭓ² = 7569.8; p < 0.001). All commonalities were above 0.3 and all items were loaded with a sufficiently high weight (> 0.4), however, four items (items 8, 13, 16 and 20) were loaded equally by several factors (see Appendix Table A2). After removing the high cross-loading items, the remaining items were subjected to confirmatory factor analysis and the model fit indicators improved significantly after the change (ꭓ2 = 527.85; Df = 97; RMSEA = 0.082; SRMR = 0.076; CFI = 0.908; TLI = 0.899). The fit indicators were further improved by removing one item (item 21) from the PE subscale and one (item 22) from the DP subscale, and by incorporating correlations between error members for four-item pairs (only between content-related items on the same factor), bringing the fit indicators of the final model version to an optimal range (ꭓ2 = 428.33; Df = 118; RMSEA = 0.071; SRMR = 0.071; CFI = 0.932; TLI = 0.913). In the case where the respondent did not respond to more than half of the items that make up each scale, the scale value was treated as missing data (Pejtersen et al., 2010). Table A3 in the appendix displays the fit indices of all the tested models. Figure 1 shows the standardised factor loadings for the final three-factor model.

Figure 1. Factor loadings for the three-factor model of MBI-HSS on the sample of Hungarian social workers.

Note: All coefficients represent standardised estimates significant at .001 level.

The resulting factors were checked for internal reliability: Cronbach α = 0.832 for the EE; α = 0.891 for the DP, and α = 0.874 for the PA. The reliability of the entire scale of 16 items was also found to be satisfactory (Cronbach α = 0.865). The results reflect that all items were loaded into the expected dimension and with a sufficiently high weight. Standardised weights range from 0.55 to 0.84 on EE, from 0.73 to 0.84 on DP, and from 0.54 to 0.83 on PA. There is a strong and positive relationship between EE and DP (r = 0.64), and a negative relationship between DP and PA (r = -0.26), as well as between EE and PA (r = -0.34). The mean scores of the subscales are as follows: 20.2 (SD: 8.32) for EE, 10.5 (SD: 4.47) for DP, and 23.8 (SD: 7.09) for PA. Social workers in our sample were experiencing, on average, moderate levels of emotional exhaustion and depersonalisation, and a low level of personal accomplishment (Maslach et al., 1996).

Linear regression models

In this study, we hypothesised that social workers working with clients in extreme poverty are more likely to burn out than those serving less poor clients. However, we had no hypothesis with regard to the different dimensions of burnout. To estimate the correlations of clients’ extreme poverty with various dimensions of burnout we used linear regression models. The models were constructed in two steps. First, the socio-demographic and working conditions variables were included in the model [(1) (3) (5) model specifications], then the variable indicating extreme poverty among clients was added to the equation [(2) (4) (6) model specifications]. The results of the linear regression models are presented in Table 1.

Table 1. Results of linear regression models of burnout (beta coefficients).

	Emotional exhaustion		Depersonalisation		Personal accomplishment
	(1)	(2)	(3)	(4)	(5)	(6)
Age cohort (ref: under 40 years)
40–49 years	−0.141 (0.097)	−0.147 (0.098)	−0.083 (0.099)	−0.086 (0.100)	0.165* (0.098)	0.194* (0.097)
50 + years	−0.271** (0.127)	−0.275** (0.127)	−0.328** (0.130)	−0.330** (0.130)	0.395*** (0.128)	0.413*** (0.126)
Experience in social work (year)	−0.003 (0.005)	−0.003 (0.005)	−0.006 (0.005)	−0.006 (0.005)	0.003 (0.005)	0.003 (0.005)
Social work degree	0.115 (0.085)	0.118 (0.085)	0.144* (0.087)	0.146* (0.087)	−0.111 (0.086)	−0.120 (0.085)
Gender (ref: female)	0.088 (0.138)	0.090 (0.139)	−0.029 (0.141)	−0.028 (0.142)	−0.097 (0.139)	−0.096 (0.138)
One−person service	0.230** (0.097)	0.229** (0.098)	0.107 (0.099)	0.106 (0.100)	−0.160 (0.098)	−0.148 (0.097)
Participated in supervision	0.106 (0.088)	0.100 (0.089)	0.136 (0.090)	0.154 (0.091)	0.098 (0.089)	0.113 (0.088)
Work overload	0.169* (0.093)	0.179* (0.094)	0.135 (0.095)	0.134 (0.091)	−0.040 (0.094)	−0.091 (0.094)
Job satisfaction: material rewards	0.232 (0.182)	0.238 (0.183)	0.262 (0.186)	0.266 (0.187)	0.377** (0.183)	0.335* (0.181)
Job satisfaction: moral esteem	−0.101 (0.109)	−0.104 (0.110)	−0.043 (0.112)	−0.044 (0.112)	0.311*** (0.110)	0.329*** (0.109)
Opportunities for professional development	−0.110 (0.118)	−0.118 (0.119)	−0.117 (0.121)	−0.120 (0.122)	0.156 (0.119)	0.173 (0.118)
Professional support	−0.450*** (0.113)	−0.443*** (0.114)	−0.361*** (0.115)	−0.359*** (0.116)	0.238** (0.114)	0.232** (0.113)
Difficulties with changing rules	0.145 (0.090)	0.148 (0.091)	0.260*** (0.092)	0.262*** (0.092)	−0.136 (0.091)	−0.149 (0.090)
Material conditions of work	−0.044 (0.086)	−0.040 (0.086)	−0.040 (0.087)	−0.039 (0.088)	0.362*** (0.086)	0.343*** (0.085)
Type of settlement (ref: capital)
Town	0.299** (0.145)	0.289** (0.138)	0.326** (0.139)	0.323** (0.141)	0.697*** (0.137)	0.704*** (0.137)
Village	0.211 (0.145)	0.192 (0.147)	0.158 (0.148)	0.150 (0.150)	0.551*** (0.146)	0.615*** (0.146)
Clients' extreme poverty		0.051 (0.092)		0.027 (0.093)		−0.332*** (0.091)
R^2	0.121	0.121	0.114	0.114	0.167	0.165
N	552	551	552	551	552	551

Note: Robust standard errors in brackets.

*p < 0.10.

**p < 0.05.

***p < 0.01.

The models are significant, and the variables included explain between 12% and 17% of the variance of the dependent variables (i.e. burnout dimensions), with the highest percentage for personal performance. The value of the VIF test indicates a lack of multicollinearity in all models (VIF < 2 and tolerance > 0.5).

According to the estimation results, of the three dimensions of burnout, only personal accomplishment shows a significant and negative relationship with clients’ extreme poverty, indicating that serving clients living in extreme poverty significantly reduces social workers’ sense of personal achievement.

Of the demographic variables, age is significant for all three burnout symptoms: social workers aged 50 or over are less likely to experience emotional exhaustion and depersonalisation, and those aged 40 or over are less likely to experience reduced personal accomplishment. This result supports that older age may be a protective factor in all three dimensions of burnout. Although the findings on the relationship between burnout and age are contradictory, several studies from abroad have found that younger social workers are the most at risk for burnout (Schwartz et al., 2008; Walters et al., 2018). A social work or related degree (‘social work degree’) has a significant and positive effect on depersonalisation: professionals with such a degree are more likely to feel depersonalised than those with other qualifications. Working in a one-person service is positively associated with emotional exhaustion: such social workers are more likely to experience emotional exhaustion than those working in larger organisations. Work overload also has a significant effect on emotional exhaustion: social workers who cannot spend enough time with their clients are more likely to be emotionally exhausted. This finding is consistent with previous research (Shier et al., 2012). Looking at the variables of recognition for work, it is evident that both material rewards and moral esteem increase the sense of personal accomplishment. These results largely confirm other Hungarian (Győri & Perpék, 2022) and international (Mor Barak et al., 2001) research findings among social workers. The effect of professional support is significant in all three models: professional recognition from the supervisor reduces emotional exhaustion and depersonalisation, while increasing the sense of personal accomplishment. Some previous research has also reported that support from immediate supervisors and organisation managers led to a reduction in burnout symptoms among social workers (Bargal & Guterman, 1996). The variable measuring difficulties with changing rules is significant and positive in the model of depersonalisation, indicating that those who find it difficult to keep up with changing rules and protocols are more likely to develop a negative attitude toward work. The findings show that better material conditions of work increase the sense of personal accomplishment. This result supports the earlier finding that social workers report greater efficiency when their physical working conditions are comfortable (Abu-Bader, 2000). Finally, the effect of settlement type is also significant in all three models of burnout. Social workers working in towns are more likely to experience emotional exhaustion and depersonalisation compared to those working in the capital. However, social workers working in towns and villages are more likely to have a higher sense of personal accomplishment than those working in the capital.

Discussion and conclusions

In this study, we investigated burnout among family and child welfare workers in Hungary in a nationally representative sample. We wanted to know whether social workers' burnout is influenced by the composition of their clients, i.e. the proportion of extremely poor people among them. Burnout not only affects the mental health of social workers but also reduces the effectiveness of social work and thus the quality of services provided to the most vulnerable. As compared to other European countries, Hungary has a relatively high number of extremely poor people for whom access to quality social services would be important, the issue is highly relevant in this country.

Working with the extremely poor requires social workers to perform a particularly large amount of emotional labour. We hypothesised that this may contribute to higher levels of burnout. Burnout was measured using the MBI which distinguishes three dimensions of burnout: emotional exhaustion, depersonalisation, and personal accomplishment. Based on preliminary research, we could not formulate a hypothesis as to which dimensions of burnout would be affected by client composition.

In our sample, the levels of emotional exhaustion and depersonalisation were very similar to those measured in other international studies among social workers, mental health professionals, and psychiatrists (Galeazzi et al., 2004; Johnson et al., 2016; Sánchez et al., 2018). However, the average score of personal accomplishment was lower compared to international experience, i.e. Hungarian social professionals working in family and child welfare services are characterised by a higher degree of performance loss.

Our results showed that working with extremely poor clients is not associated with higher emotional exhaustion and higher depersonalisation than working with less poor clients. However, social workers’ sense of personal accomplishment was found to be significantly lower when attending to extremely poor families. Thus, exposure to extreme poverty does indeed increase burnout among service providers, namely through low perceived personal accomplishment. This research finding helps to better understand which aspects of working with the extremely poor are the most stressful for social workers.

However, an alternative explanation for our findings is also possible, which is related to the criticisms of the applicability of the MBI. Although this is perhaps the most widely used measure of burnout to date, there have been many criticisms of its use. These include questioning whether personal accomplishment (or professional efficacy) is a component of the burnout syndrome. The arguments are both technical (PA behaves differently in models than the other two dimensions of burnout) and conceptual (reduced PA is seen as a consequence rather than an integral part of burnout) (Schaufeli & Taris, 2005). Schaufeli et al. (2020), along with several others, therefore propose the removal of PA from the concept of burnout.

Our results support the argument that PA is largely independent of the other two components. Moreover, they raise the possibility that the low sense of personal accomplishment is not necessarily driven by negative self-evaluation, dissatisfaction, or lack of competence, in line with the original concept of burnout. In our case, the low perceived self-efficacy of social workers working with the very poor may be a realistic assessment of their potential to make a positive impact on the lives of their clients. Attending to clients living in extreme poverty can certainly be emotionally demanding. Social workers who witness their clients' poverty-related difficulties on a daily basis can feel empathetic to their clients' struggles. At the same time, they may be frustrated by institutional abuse, i.e. the inadequacy of available resources in relation to the scale and nature of the problem, which prevents them from providing an adequate service to their clients. Such situations may be experienced as a failure of professional performance, even though the solution to the complex problem of extreme poverty does not depend on them alone. Low PA may therefore be driven more by feelings of powerlessness and lack of control over their work, resulting from a lack of resources to tackle the systemic problem of extreme poverty.

If burnout is interpreted in terms of the two dimensions of emotional exhaustion and depersonalisation, our results show that the burnout of social workers who work with the extremely poor is not significantly increased by their perception of low professional efficacy. However, it is possible that by using a different concept of burnout we would obtain different results. Schaufeli et al. (2020), for example, argue that, in addition to emotional exhaustion, we should also consider emotional impairment (reduced functional capacity for emotional regulation) as a new dimension of burnout. Such an extension of the concept of burnout syndrome may be worth testing in studies of social workers.

There should be further research into how social workers working with the extremely poor can avoid chronic stress and burnout. As a coping strategy, they may focus more on aspects of helping clients that are less emotionally demanding for them (e.g. spending more time on office work and less on face-to-face contact), or on tasks over which they have more control (e.g. helping people access benefits rather than providing them more complex support). It is also possible that they simply invest less energy in the most vulnerable clients, i.e. those who need their support the most. These coping strategies might be beneficial to the social worker's mental and physical health but may result in a lower quality of work.

Our study has some limitations. First, as the applied database is cross-sectional, our results do not allow us to show causal relationships but instead suggest plausible explanations. Future longitudinal studies would be needed to confirm the causal nature of our results. Second, it is worth noting that our research took an organisational rather than a clinical approach to burnout, so higher scores on the burnout questionnaire do not necessarily indicate higher levels of clinical burnout (van Dam, 2021). But even if we cannot interpret our findings in terms of clinical burnout, it is important to address the significant consequences of chronic stress that professionals attending to the extremely poor face, which exposes them to various mental and physical illnesses in the long term. As extreme poverty is present to some extent in all EU member states, social workers should be better prepared to cope with the increased emotional strain it brings, so that it does not lead to mental health problems.

Acknowledgments

Ildikó Husz acknowledges financial support from the ‘Professional Support for Integrated Child Opportunities Programmes’ under Grant No. HRDOP-1.4.1–15; Ágnes Győri acknowledges financial support from the National Research, Development and Innovation Fund under Grant No. FK 138315.

Disclosure statement

The authors report that there are no competing interests to declare.

Additional information

Notes on contributors

Marianna Kopasz

Marianna Kopasz is a Senior Research Fellow at the Institute for Political Science, Centre for Social Sciences. Her primary research interests include public policy-making, child protection, social services, and social workers practice. She is currently leading a research project on the functioning of the child protection referral system in schools.

Ágnes Győri

Ágnes Győri, is Senior Research Fellow at the HUN-REN Centre for Social Sciences, Hungarian Academy of Sciences Centre of Excellence. She holds a PhD in Sociology from the Corvinus University of Budapest. She focuses in her research and publications on mental and physical health, social and health inequalities, mental health and interpersonal networks, burnout and work environment of social professionals. She is the principal investigator of a Hungarian NRDI FUND project entitled ‘Professional working conditions, mental health and mobility in the social sector’.

Ildikó Husz

Ildikó Husz is a Senior Research Fellow at the HUN-REN Centre for Social Sciences, where she currently heads the Child Opportunities Research Group. She is also a Senior Research Fellow at the Corvinus University of Budapest. Her research focuses on child poverty, ethnic inequalities, and spatial inequalities in social services. She is the research leader of the project 'Professional Support for Integrated Child Opportunities Programmes' funded by the European Social Fund and co-financed by the Hungarian Government.

Márton Medgyesi

Márton Medgyesi, PhD Sociology, is Senior Research Fellow at Corvinus Institute for Advanced Studies and the Child Opportunities Research Group, Centre for Social Sciences in Budapest. His main research interests are the empirical analysis of income distributions and the redistributive effect of welfare benefits and social services. He worked as principal investigator of several research projects on behalf of the European Commission, Eurofound, Global Development Network and the Hungarian Scientific Research Fund.

Notes

1 In the sample, only 42.8% of the social workers have a mobile phone provided by their workplace for contacting clients, so we believe that this variable is an appropriate proxy for material working conditions in Hungary.

Appendix

Table A1. Descriptive statistics of variables included in the analysis.

Variable	Obs	Mean	SD	Min	Max
Emotional exhaustion	574	0.022	0.995	−1.30	3.55
Depersonalisation	574	−0.024	1.010	−1.11	3.39
Personal accomplishment	574	−0.025	1.042	−2.58	1.79
Age: Under 40 years	600	0.501	0.500	0	1
Age: 40–49 years	600	0.312	0.464	0	1
Age: 50 + years	600	0.192	0.390	0	1
Experience of social work	600	9.447	8.283	0	44
Social work degree	600	0.627	0.486	0	1
Gender (male)	600	0.110	0.307	0	1
One-person service	600	0.253	0.435	0	1
Participated in supervision	595	0.388	0.487	0	1
Work overload	598	0.659	0.474	0	1
Job satisfaction: material rewards	597	0.347	0.475	0	1
Job satisfaction: moral esteem	598	0.806	0.401	0	1
Opportunities for professional development	596	0.381	0.485	0	1
Professional support	597	0.472	0.499	0	1
Difficulties with changing rules	596	0.563	0.497	0	1
Material conditions of work	600	0.568	0.495	0	1
Capital	600	0.118	0.325	0	1
Town	600	0.516	0.500	0	1
Village	600	0.368	0.481	0	1
Client's extreme poverty	599	0.716	0.455	0	1

Table A2. Factor loadings of the exploratory factor analysis of the MBI-HSS on the sample of Hungarian social workers.

	Exploratory factor analysis
	Factor 1	Factor 2	Factor 3	h^2
MBI-HSS 5 item	0.8747			0.7783
MBI-HSS 15 item	0.8454			0.7172
MBI-HSS 10 item	0.7783		0.3333	0.7365
MBI-HSS 11 item	0.6614			0.6851
MBI-HSS 16 item	0.6268		0.4160	0.6116
MBI-HSS 20 item	0.6287		0.5701	0.6880
MBI-HSS 8 item	0.6268		0.5701	0.7476
MBI-HSS 22 item	0.5859			0.4203
MBI-HSS 19 item		0.8126		0.6738
MBI-HSS 17 item		0.7842		0.6615
MBI-HSS 12 item		0.7773	−0.3305	0.7140
MBI-HSS 7 item		0.7680		0.6388
MBI-HSS 21 item	−0.3102	0.7446		0.6616
MBI-HSS 9 item		0.7227		0.5411
MBI-HSS 4 item	−0.3146	0.6949		0.6522
MBI-HSS 18 item		0.6757		0.6636
MBI-HSS 2 item			0.8223	0.6988
MBI-HSS 1 item			0.7766	0.6503
MBI-HSS 6 item	0.3553		0.7294	0.6708
MBI-HSS 3 item	0.3298		0.6753	0.5841
MBI-HSS 13 item	0.4831		0.6663	0.6870
MBI-HSS 14 item			0.5160	0.3904
Eigenvalues	8.2198	3.3908	2.0937

Note: Factor loadings are derived using a varimax rotation with Kaiser Normalisation. Only factor loadings > 0.30 are shown. Bold face is factor loadings relevant to the item > 0.40, h² communality.

Table A3. Fit indexes for all tested models of MBI-HSS among Hungarian social workers.

Models	ꭓ^2	df	RMSEA [90% CI]	SRMR	BIC	CFI	TLI
Model 1 (22 items): the original Maslach’ model	1400.8	206	0.106 [0.101–0.111]	0.081	36892.97	0.841	0.820
Model 2 (19 items)*: similar to Model 1, with items 14, 21 and 22 deleted and error covariances between various items	774.68	145	0.091 [0.085–0.097]	0.069	31695.89	0.906	0.889
Model 3 (18 items): similar to Model 1, with items 8, 13, 16 and 20 deleted	902.43	132	0.088 [0.081–0.095]	0.084	30994.86	0.898	0.872
Model 4 (18 items): similar to Model 1, with items 8, 13, 16 and 20 deleted and error covariances between various items	771.81	128	0.083 [0.078–0.089]	0.080	30889.29	0.903	0.897
Model 5 (16 items): similar to Model 1, with items 8, 13, 16, 20, 21 and 22 deleted	527.85	97	0.082 [0.076–0.088]	0.076	27778.11	0.908	0.899
Model 6 (16 items): similar to Model 1, with items 8, 13, 16, 20, 21 and 22 deleted and error covariances between various items	428.33	118	0.071 [0.064–0.078]	0.071	26188.25	0.932	0.913

Note: Model 2 was estimated as documented by the MBI-HSS validation study in Hungary (Ádám & Mészáros, 2012).