Sage Journals: Discover world-class research

Abstract

This article addresses a gap in comparative research regarding the role of organisational factors in shaping occupational health and safety (OHS) outcomes when using temporary agency work (TAW). Using the Economic Pressures–Work Disorganisation–Regulatory Failure (PDR) framework, the article applies a comparative case study approach to analyse OHS outcomes in two sites of one multinational company. The findings reveal how differences in organisational practices embedded in different institutional contexts shape TAW use, influencing exposure to work-related injuries and exhaustion. These effects are particularly evident in the context of workforce segmentation, where trade-offs and unintended consequences emerge for the user firm.

Keywords

Comparative case study exhaustion OHS PDR model temporary agency work work injury

Introduction

Temporary agency work (TAW) has been established as one of the most problematic employment types in terms of job quality, standing out even among other forms of precarious work due to particular challenges in occupational health and safety (OHS) (Håkansson et al., 2013; Underhill and Quinlan, 2011). Its triangular combination of employment and business relationships allows for the outsourcing of not only ‘bad jobs’ but also their negative externalities in terms of health outcomes (Lewchuk et al., 2008; Lippel et al., 2011, Underhill and Quinlan, 2011). Therefore, agency workers are commonly found to be at high risk of work accidents (Arbetsmiljöverket, 2013; Hintikka, 2011), though associations between TAW and mental health outcomes are inconclusive (Hünefeld et al., 2020). Comparative research in this area is characterised by heterogeneity in national contexts, sectors, firms and jobs, as well as different inclusion criteria and operationalisations of contract types and confounding variables (Håkansson et al., 2013, Hünefeld et al., 2020).

Understanding job quality as multidimensional, contextual and relational (Munoz de Bustillo et al., 2009) requires a deeper exploration of the associations between TAW and OHS as well as sensitivity to organisational contexts, and the regulatory environments that govern employment relationships (Hünefeld et al., 2020; Kim et al., 2012; Quinlan, 2015). Against this background, the lack of cross-national comparative research examining the contextual mechanisms behind employment-based inequalities is notable, particularly given the potential for preventive interventions through employment policy (Leka et al., 2010).

This case study draws on unique firm-level data that allow for a comparison of employee experiences and organisational practices within one multinational company (MNC) operating under two regulatory regimes in the EU: Sweden and Poland. Using a mixed-method design, qualitative data examine how employers adapt and embed TAW use into local institutional contexts. Building on the Economic Pressures–Work Disorganisation–Regulatory Failure (PDR) framework of workplace health (Quinlan and Bohle, 2009), the article also uses employee-level survey data to explore the heterogeneity in OHS outcomes. Two interrelated research questions are addressed: first, whether agency workers and user firm employees performing the same jobs in these two cases experience different levels of work-related injuries and exhaustion; and second, what contextual factors explain these OHS outcomes.

Contextual drivers in OHS: PDR model of workplace health

As work-related health outcomes develop through a complex interplay of contextual factors that shape employment relationships, organisational practices, and individual vulnerabilities (Dellve et al., 2008), a deeper understanding of the mechanisms behind workplace risks and opportunities for their mitigation requires expanding our theoretical frameworks to encompass a wide range of factors not only at the individual level but also at the organisational level and within broader contexts (Nielsen et al., 2010). The PDR model (Quinlan and Bohle, 2009; Quinlan et al., 2001) incorporates three key dimensions of impact on workplace health: economic and reward pressures (P), work disorganisation (D) and regulatory failure (R). Unlike other workplace health models, it focuses on contextual mechanisms, with elements of its three dimensions expected to exacerbate existing risks and contribute to the emergence of new ones across all forms of work, but particularly in precarious employment. The model’s flexibility allows for the examination of factors conceptualised at both the individual and organisational levels and their associations with a range of OHS outcomes. To date, the model has mainly been applied in the Australian context, although it is recognised as a promising framework for advancing the understanding of associations between TAW and work-related health in Europe also (Hünefeld et al., 2020).

Economic pressures

The first dimension of the PDR model, economic pressures, aligns with the concept of employment strain (Lewchuk et al., 2008) and is operationalised through pressures and insecurities embedded in organisational reward systems and labour market uncertainties. These factors lead workers to experience financial strain and promote work intensification (Bohle et al., 2015). The insecurity associated with short-term contracts and assignments in TAW exposes workers to vulnerabilities driven by employer demands, which can undermine their ability and motivation to engage in safe behaviours at work (Strauss-Raats, 2019).

Previous research has found a positive association between job or employment insecurity and poor psychological and physical health (Ferrie et al., 2008; Sverke et al., 2002), as well as safety performance (Sverke et al., 2019) and work accidents (Probst and Brubaker, 2001). Agency workers tend to experience higher levels of insecurity than workers in standard employment (Dütsch, 2011; Wagenaar et al., 2012), even when they are on open-ended agency contracts (Håkansson et al., 2012). In the Swedish context, Håkansson and Isidorsson (2016) found that agency workers who were less satisfied with job security had a higher risk of work injury. Probst et al. (2018) also found that insecurity moderates the association between precarious work and poor safety behaviour. However, the moderating role of insecurity in linking TAW to mental health requires further research (Hünefeld et al., 2020).

Work disorganisation

Work disorganisation captures the inefficiencies in organising work and OHS management that result from the use of atypical labour. The triangular relationship that governs TAW use blurs employer responsibilities, hinders the maintenance and communication of workplace regulations, fractures work communities, and risks mismatches between workers and tasks (Underhill and Quinlan, 2011; Ward et al., 2001). This aligns with a systems perspective which considers workplace OHS as a dynamic process, in which accidents arise from uncontrolled disruptions and interactions within the work organisational system (Leveson, 2015), or from delays in feedback from system operation (Casey et al., 2017).

Work disorganisation is evident in uneven hazard exposure or the exclusion of TAW in workplace risk assessments (Strauss-Raats, 2019), role uncertainty (Håkansson and Isidorsson, 2016) and insufficient social support (Byoung-Hoon and Frenkel, 2004; Wilkin et al., 2018). Given the inherent interconnectedness of organisational structures, TAW use also involves ‘spillover’ risks, reflected in strained social relationships (Davis-Blake et al., 2003; Ward et al., 2001), perceptions of work intensification (Holst et al., 2010), and insecurity and anxiety within the user firm (Bryson, 2013).

Regulatory failure

Finally, the PDR model identifies regulatory failure as a distinct mechanism, which contributes to poor OHS outcomes through issues in the application, monitoring and enforcement of labour standards (Quinlan and Bohle, 2009; Quinlan et al., 2001). Within legal frameworks primarily designed for standard employment (Vosko, 2011), most regulatory regimes have gaps in addressing the workplace issues raised by the use of TAW. For example, it has been well established that a precarious position at the labour market is a hindrance for meaningfully participating in the OHS management systems that are built on the assumption of workers’ active role raising safety concerns and dialogue between employers and employees (Aronsson, 1999).

Contrasting regulatory contexts: Sweden and Poland

The two cases in this study represent different regulatory regimes for TAW in the EU. Sweden, with its low state interference in employment regulation and limited statutory restrictions on TAW use, is described as a ‘liberalised’ regime (Sartori, 2016). The Swedish model has also been characterised as one of ‘organised decentralisation’ to denote the strong union presence at the workplace level coupled with decades of retrenchment policies targeted at reorganising and slimming down state funding for OHS services (Thörnquist, 2008). In practice, this means that while the Swedish state has defunded OHS in recent decades, trade unions maintain extensive responsibilities in coordinating and implementing OHS at the workplace and sectoral levels. Furthermore, TAW employment is protected by sectoral-level collective bargaining agreements, aligning it with standard employment (Coe et al., 2009) and sharing the norm of open-ended contracts prevalent in the national labour market (Ahlberg, 2008). Agency workers can join sectoral unions, and the blue-collar collective agreement provides for guaranteed pay between assignments, hourly wages based on comparable groups at the user firm, and tenure-based dismissal periods ranging from two weeks to six months. About 2.6% of Swedish employees worked through temporary agency contracts in 2023 (Eurostat, 2025).

In contrast, the Polish ‘regulated’ TAW regime (Sartori, 2016) relies on statutory law to restrict TAW use and prevent its (mis)use, with social partners’ involvement being limited and primarily active at the firm level. Employee safety representatives play a consultative role and can be elected either by members of trade unions or by employees if no union operates in the company (EU-OSHA, 2022). TAW is permitted for specific needs, such as temporary increases in workload, but is prohibited for certain hazardous jobs. An estimated 0.5% of Polish employees were employed through temporary agency contracts in 2023 (Eurostat, 2025). While the law mandates equal pay and benefits, it restricts consecutive assignments to a maximum of 18 months within a 36-month period, with dismissal periods ranging from three days to two weeks. Union representation for agency workers is limited (Mrozowicki and Maciejewska, 2016). Although the two major Polish trade union confederations have made attempts to organise and represent temporary agency workers, these efforts have had limited success (Marino et al., 2019).

In both countries, the temporary work agency is recognised as the employer of agency workers for OHS purposes. In Poland, the law clearly divides OHS obligations between the user firm and the agency, whereas in Sweden, OHS responsibilities are shared between the agency and the user firm (Håkansson et al., 2013). Agency workers employed on employment contracts are covered by general social security guarantees in both countries (Håkansson et al., 2013; Mrozowicki and Maciejewska, 2016). In both contexts, firm-level union safety representatives represent all workers in OHS matters but not in general employment matters.

The European Labour Force Survey reports that the work accident rate in Sweden is nearly five times higher than in Poland, although Polish rates are likely affected by underreporting (EU-OHSA, 2017). The Swedish Work Environment Authority estimates that the accident rate for TAW is double that of other sectors (Arbetsmiljöverket, 2013). Comparable reliable data for Poland are unavailable, but it is likely that TAW accident rates are higher than in general employment (Mrozowicki and Maciejewska, 2016).

Materials and methods

Research design, case organisation and participants

This article is part of a mixed-method comparative case study that draws on a unique set of qualitative and quantitative data, with the main inferences based on a quantitative employee survey. The comparisons follow a most-similar case design (George and Bennett, 2005), where there is similarity at the organisational level but divergence in national regulatory contexts. The Swedish context serves as a ‘critical case’ for TAW economic security, providing a high degree of protection for agency workers from economic pressures, while the Polish context represents a case of relative insecurity. The two cases of TAW use involve comparable units within the same multinational company (MNC), which follows the same global ‘high-road’ HR policies and utilises the same production system, with agency workers and user firm employees performing the same jobs.

Qualitative and quantitative data were integrated sequentially (Seawright, 2016). The first step involved reviewing available organisational policies and guidelines related to TAW and OHS, followed by 14 semi-structured expert interviews (Table 1) with employer and trade union representatives. Interviewees were selected based on their responsibilities in OHS and TAW. The interviews were conducted at the two production sites, allowing for complementary on-site observations. Data from qualitative sources informed the employee survey design and were used again to contextualise and interpret survey results.

Table 1.

List of interviewees and interview languages in the two organisational cases.

	Polish case	Swedish case
HR policy manager/specialist	3 (English)	1 (English)
Line/production manager	2 (English)	1 (English)
Head of local trade union	1 (Polish, interpreted into English)	1 (English)
Trade union head safety rep	1 (Polish, interpreted into English)	1 (English)
Company safety rep	1 (English)	No comparable function
Sectoral regional trade union rep	No comparable function	2 (Swedish/English)
Total interviewed	8	6

In the Swedish case, questionnaires were distributed and collected by research assistants during department meetings to all production workers. Of the 209 questionnaires distributed, 202 were returned (RR = 97%). Excluding those on probationary contracts resulted in 176 valid questionnaires, 58 (31%) of which were by agency workers. In the Polish case, questionnaires were distributed during working hours to be completed at home and returned to locked boxes at the workplace. A total of 498 responses were collected from the 1,215 questionnaires distributed (RR = 40.1%). Agency workers returned 49 (11%) of the surveys. We excluded 23 questionnaires from the analysis due to missing information on contract type. The share of agency workers among respondents is close to the overall share of agency workforce in both cases (ca. 30% in Swedish case and 15% in Polish case, according to the company-provided data).

Survey variables

The first dependent variable captures the experience of a work-related injury in the past 12 months, coded as 1 if the respondent responds ‘Yes’ to the question ‘Please indicate, if during the past 12 months you have had an accident at work that did not require taking time off work’ or ‘an accident at work that resulted in time off from work’ and 0 otherwise. The second dependent variable, work-related exhaustion, reflects experiences of physical, affective and cognitive strain, operationalised through an eight-item scale from the Oldenburg Burnout Inventory (OLBI) (Demerouti et al., 2003). The OLBI scale includes both positively and negatively worded items, scored on a four-point Likert scale ranging from ‘Strongly agree’ to ‘Strongly disagree’. Item ratings were averaged into a single score (1–4), with a higher score indicating greater exhaustion. Cronbach’s alpha values demonstrated a high level of reliability, with a value of .82 for the Polish case and .83 for the Swedish case.

Contract type, as an independent variable, was coded as 1 for respondents employed on a TAW contract and 0 for those on an MNC contract. Low tenure was controlled for, with a value of 1 if the worker had been with the MNC for less than a year and 0 otherwise. Age was measured in years.

The PDR model was incorporated into two sets of variables. Economic pressures were captured through perceived work-related insecurities, using affective worry rather than cognitive evaluation to better align with the concept of pressure (Berglund et al., 2014). The different dimensions of pressure were assessed with follow-up questions to the prompt, ‘How much are you worried about the following: Becoming unemployed (job insecurity); A reduction in my salary that would make it difficult to make ends meet (income insecurity); Finding another job that is as good or better if I become unemployed (employment insecurity); Being transferred to another job against my will (assignment insecurity).’ Response options were provided on a five-point scale, ranging from ‘To a very large extent’ to ‘To a very small extent’.

Work disorganisation was measured using six variables that assessed experiences of organisational OHS practices. Involvement in risk assessment was evaluated with the question: ‘Has someone at this workplace asked you about what kind of health and safety risks you see in your current job?’ Use of occupational health services was assessed by asking: ‘Have you ever discussed your current job here in the MNC with a doctor or occupational health service to determine if it is suitable for you and does not harm your health?’ Both were coded as 1 for ‘Yes’ and 0 otherwise. Perceived supervisory support at the MNC was measured using a four-item self-constructed scale that captured dimensions of emotional support (‘Considers how I am feeling at work’, ‘Is easy to talk to when I need to discuss important personal matters’), informational support (‘Gives me advice on how to handle things when I need it’) and instrumental support (‘Helps me out when things get tough’). A five-point response scale was used in Poland and a four-point scale in Sweden, ranging from ‘Always’ to ‘Never’ and ‘Often’ to ‘Never’, respectively. Cronbach’s alpha values of .93 indicated high reliability in both cases. Work intensity was measured by the single item, ‘Do you have to work very fast?’ from COPSOQ II (Pejtersen et al., 2010), with responses ranging from ‘Always’ to ‘Never’ in the Polish case and from ‘Often’ to ‘Never’ in the Swedish case. Two items from the European Working Conditions Survey captured exposure to chemical and physiological hazards (‘Breathing in smoke, fumes such as welding or exhaust fumes, powder or dust, etc.’, ‘Tiring or painful positions, repetitive hand or arm movements’), with responses on a six-point scale ranging from ‘All of the time’ to ‘Never’. Responses for continuous independent variables were rescaled to a 0–100 range.

Empirical approach

Data analyses began with checks for coding errors, missing values and assumptions for multivariate analysis. The variable Age had a relatively high proportion of missing data (13% and 22% for user firm employees, and 6% and 7% for agency workers in the Polish and Swedish cases, respectively). To preserve sample size and reduce potential bias from complete case analysis (Little and Rubin, 2019), mean imputation for Age was applied based on contract type and country.

ANOVA and chi-square tests were used to explore differences between the two employee groups across all variables within cases. Stepwise logistic regression and OLS regression were employed to examine associations between injuries and work-related exhaustion with agency contract status, economic pressures and work disorganisation. The initial models included only the agency contract variable, followed by controls for low tenure and age. Full models incorporated variables capturing economic pressures and work disorganisation. To account for the nonlinear relationship between age and exhaustion, a squared term for age was included in the OLS models, which improved model fit. Variance inflation factor (VIF) analysis indicated that collinearity was not present. Considering missing data and comparability, observations in samples across stepwise models were restricted to those included in the final models (Williams, 2015).

Due to the low sample size, interaction effects between the agency contract and other independent variables were explored by adding interaction terms separately to the final models, one at a time. To aid in the interpretation of interactions, marginal effects at representative values were estimated, holding other variables at their average value (Mize, 2019; Williams, 2012) for independent variables with a significant interaction effect at the p=.10 level. Given the differences in sample sizes and the case study’s focus on quantitative inferences within the study populations, the interpretation of results relied on the relative magnitude and direction of effects, using 95% confidence intervals (Greenland et al., 2016; Wasserstein and Lazar, 2016). Qualitative cross-case comparisons were visually supported by a combined plot of regression results (Jann, 2014)

Results

Qualitative data: Organisational practices of TAW use and OHS

Reflecting a shared global HR policy, employer representatives in both cases described in interviews that while the MNC was the employer of choice in their respective areas, access to blue-collar jobs was mediated by work agencies. Although there was a wish to recruit new permanent employees, employer representatives described being constrained by MNC headcount limits combined with low voluntary turnover and strong employment protection legislation. Interviews and documents revealed that, while both cases shared flexibility goals in their use of TAW, the implementation of TAW was adapted to local contexts.

Poland: TAW as a flexibility buffer

In the Polish case, work was organised along core–periphery lines, with agency workers being used for numerical flexibility in low-skilled tasks and excluded from training and production meetings. Employer representatives described this approach as a response to legal restrictions that allow TAW contracting only for a limited duration, leading to a lack of incentives to invest in the temporary workforce. Since consecutive contracts were not restricted, the common practice was to renew agency contracts weekly or monthly based on production needs. According to an employer representative, the use of TAW left user firm employees with more complex tasks, quality surveillance and training obligations.

The union representatives at the MNC described their role in representing all workers in OHS matters but also actively bargaining over TAW conditions to maintain overall job quality. Local agreements enforced the use of user firm clothing and PPE and established a 15% cap on the agency workforce. Upon reaching this cap, some agency workers were offered temporary contracts with the MNC based on supervisors’ assessments of motivation and performance; however, permanent direct employment was not an option. Agency workers’ pay was determined by comparable jobs, accounting for tenure and skills, which resulted in comparatively lower wages than those of user firm employees. Contact between workers and the agency was described as very limited, with most supervisory responsibilities falling on the user firm. While agency workers participated in assessing physical risks alongside user firm employees, they were excluded from the employee survey capturing psychosocial work environment. Accidents were reported to user firm supervisors, and work agencies were not actively engaged with the user firm in OHS management.

Sweden: Strategic use of TAW

In the Swedish case, TAW was strategically managed through long-term cooperation with a single agency, with a TAW penetration rate of around 30%. Some semi-permanent agency workers had been with the MNC for years, and TAW was fully integrated into the user firm’s work organisation. Agency workers contributed to both numerical and functional flexibility, performing the same jobs and participating in competence development that enhanced their company-specific skills. Employer representatives noted that user firm employees were less inclined to embrace functional flexibility as due to concerns about managing their workload and associated risks.

Agency employees were employed on open-ended contracts. However, the relative job security provided by these contracts came at the cost of assignment insecurity, with the possibility of being temporarily reassigned to other users at short notice. While not a daily threat, this option affected wage security, as wages were based on the workplace’s hourly average, and the MNC offered higher wages. According to an employer representative, the functional flexibility contributed by agency workers, combined with the organisational need to maintain firm-specific competences, led to recent recruitment of some agency workers into permanent positions.

The local union was not actively involved in negotiations around TAW, but the MNC adhered to the sectoral collective agreement. Local safety representatives represented all employees in OHS matters, and unionised agency workers could also receive support from regional union representatives. In addition to a supervisor at the user firm, agency workers had an ongoing supervisory relationship with an agency coordinator. Interviewees had no knowledge of a risk assessment process prior to TAW introduction nor was there any cooperation between the agency and the company in connection to this process – it was assumed that the agency fulfilled their own legal obligations internally. MNC internal risk assessments were conducted in expert groups in consultation with employees; however, according to employee representatives, agency workers were excluded due to their temporary roles. Also, the access to occupational health services was divided by employment – the agency had a contract with a service provider in a nearby town, whereas on-site services were available for user firm employees.

Survey results

While the proportion of employees who had experienced a work injury in the past year was higher among agency workers in both cases, the mean values for work-related exhaustion show the opposite pattern, with user firm employees reporting higher exhaustion levels than agency workers (Table 2). Descriptive statistics also show that agency workers were on average younger than user firm employees and had significantly lower tenure within the user firm in both cases. Notably, agency workers experienced much higher levels of insecurity compared to user firm employees across all four dimensions in the Swedish case, whereas in the Polish context, contract groups differed significantly only in job insecurity. User firm employees were more often involved in risk assessments when compared to agency workers in the Polish case, while the pattern was the opposite in the Swedish case, with higher involvement among agency workers. Occupational health service consultation was notably lacking for Swedish agency employees, with only one worker reporting a consultation compared to 40% of directly-hired surveyed employees. Agency workers reported higher managerial support from user firm supervisors than their user firm colleagues in both cases. Exposure to workplace hazards was similar for both contract groups in Sweden but not in Poland, where agency workers reported higher exposure to chemical risks but also a lower pace of work.

Table 2.

Description of samples by contract type in the Swedish and Polish case.

	Polish case			Swedish case
	User employees	Agency workers	Sig	User employees	Agency workers	Sig.
n	419 (89%)	49 (11%)		118(63%)	58(31%)
Age (M, SD)	44.9 (10.2)	36.9 (12.3)	t=4.94 p=.001	44.9 (10.7)	28.4 (6.7)	t=10.14 p=.001
Tenure under 1 year	3%	48%	chi ² =124.18 p=.001	4%	38%	chi ² =35.16 p=.001
Work injury	8%	20%	chi ² =8.26 p=.004	27%	36%	chi ² =1.53 p=.217
Work-related exhaustion (scale 1–4, M, SD)	2.48 (.55)	2.30 (.67)	t=1.95 p=.05	2.54 (.53)	2.39 (.55)	t=1.72 p=.087
Income insecurity^{a º}	29%	29%w	chi ² =1.11 p=.292	9%	45%	chi ² =9.60 p=.002
Job insecurity^{a º}	25%	76%	chi ² =10.36 p=.001	10%	72%	chi ² =25.36 p=.001
Employment insecurity^{a º}	75%	61%	chi ² =.456 p=.500	38%	52%	chi ² =4.37 p=.037
Assignment insecurity^{a º}	43%	59%	chi ² =.052 p=.819	36%	71%	chi ² =22.96 p=.001
Involvement in risk assessment	78%	65%	chi ² =4.63 p=.031	57%	70%	chi ² =2.84 p=.092
Occupational health services	72%	84%	chi ² =3.07 p=.08	40%	2%	chi ² =28.55 p=.001
Supervisor support^º (scale 0–100, M, SD)	54.45 (24.35)	65.20 (25.06)	t=−2.88 p=.005	60.76 (27.32)	67.94 (25.64)	t=−2.88 p=.005
Work intensity^º (scale 0–100, M, SD)	66.34 (23.02)	59.18 (22.65)	t=2.06 p=.04	83.19 (20.69)	83.98 (21.83)	t=−.236 p=.814
Chemical hazard^º exposure (scale 0–100, M, SD)	33.78 (37.32)	40.96 (39.16)	t=−1.24 p=.215	16.06 (29.52)	13.21 (25.61)	t=.602 p=.548
Physiological hazard exposure^º (scale 0–100, M, SD)	50.85 (36.45)	51.02 (36.43)	t=−.031 p=.975	60.55 (33.08)	58.48 (37.42)	t=.366 p=.715

For more intuitive interpretation, insecurity dimensions show proportion of workers reporting worry to a high or very high degree.

Unstandardised values are shown for study variables in the table, statistical analysis is based on standardized values.

The regression analysis presented in Tables 3 and 4 examines whether existing differences in injury and exhaustion exposure can be explained by factors other than contract type. Logistic models (Models 1–6) in Table 3 use odds ratios, while linear models (Models 7–12) in Table 4 present standardised coefficients for continuous variables (with Age as an unstandardised variable). Table 3 indicates that, in the Polish case, the odds of experiencing an injury in the past year were nearly four times higher for agency workers than for user firm employees. This effect is reduced when controlling for age and low tenure and further decreases when accounting for economic pressures and work disorganisation. In Model 3, the confidence intervals include 1, but their bounds suggest that a positive association between an agency contract and the odds of an injury remains plausible.¹.

Table 3.

Results from logistic regression. Estimated odds ratios (OR) and 95% confidence intervals (CI) from a logistic regression of work-related injury.

Dependent variable = occurrence of work injury during the past year
	Polish case			Swedish case
	Model 1TAW	Model 2Controls	Model 3PDR	Model 4TAW	Model 5Controls	Model 6PDR
Agency worker(User firm employee=0)	3.76(1.55, 9.10)	3.32(1.10, 9.98)	2.93(0.81, 10.54)	1.59(0.78, 3.25)	1.26(0.48, 3.29)	2.40(0.66, 8.76)
Tenure < 1year(Tenure over 1 year=0)		0.59(0.14, 2.46)	0.93(0.19, 4.51)		1.61(0.57, 4.56)	2.46(0.72, 8.38)
Age (years)		0.95(0.91, 0.99)	0.93(0.89, 0.99)		1.00(0.96, 1.04)	1.00(0.96, 1.06)
Income insecurity			1.02(0.61, 1.71)			1.24(0.73, 2.10)
Job insecurity			0.63(0.33, 1.21)			1.00(0.59, 1.71)
Employment insecurity			1.35(0.75, 2.44)			1.30(0.84, 2.02)
Assignment insecurity			1.58(0.89, 2.80)			0.57(0.32, 1.04)
Involvement in risk assessment (No involvement=0)			0.32(0.13, 0.77)			0.95(0.40, 2.27)
Occupational health services (No contact=0)			1.03(0.37, 2.85)			1.53(0.58, 4.07)
Supervisorsupport			0.80(0.50, 1.27)			0.95(0.61, 1.47)
High pace			1.81(1.07, 3.06)			1.38(0.80, 2.36)
Chemical hazard exposure			1.78(1.09, 2.92)			1.70(1.04, 2.75)
Physiological hazard exposure			0.72(0.42, 1.23)			2.14(1.32, 3.49)
Pseudo R²	0.04	0.06	0.19	0.009	0.014	0.16
Observations	409	409	409	154	154	154

Confidence intervals in parentheses.

Table 4.

Results of an OLS regression on work-related exhaustion showing estimated coefficients and 95% confidence intervals.

Linear models: DV work-related exhaustion
	Polish case			Swedish case
	Model 7TAW	Model 8Controls	Model 9PDR	Model 10TAW	Model 11Controls	Model 12PDR
Agency worker(User firm employee=0)	–0.25(–0.42, –0.07)	–0.18(–0.41, 0.05)	–0.23(–0.43, –0.03)	–0.17(–0.34, 0.02)	0.05(–0.19, 0.29)	0.03(–0.19, 0.24)
Tenure < 1year (Tenure over 1 year=0)		–0.09(–0.34, 0.17)	0.02(–0.20, 0.24)		–0.25(–0.52, –0.02)	–0.07(–0.23, 0.19)
Age (years)		0.03(–0.01, 0.07)	–0.004(–0.03, 0.04)		0.04(–0.01, 0.09)	0.07(0.03, 0.11)
Age squared		–0.001(–0.001, 0.000)	–0.0002(–0.001,0.0002)		–0.0004(–0.001, 0.0002)	–0.001(–0.001, –0.0003)
Income insecurity^a			0.08 (0.002, 0.15)			–0.07 (–0.16, 0.02)
Job insecurity^a			0.08 (0.02, 0.14)			–0.06 (–0.14, 0.03)
Employment insecurity^a			–0.08(–0.15, 0.01)			0.06(–0.01, 0.13)
Assignment insecurity^a			0.09(0.03,0.15)			0.16(0.07, 0.26)
Involvement in risk assessment (Not involved=0)			–0.12 (–0.24, –0.01)			–0.002 (–0.14, 0.14)
Occupational health services(No contact=0)			–0.10(–0.20, 0.01)			0.07(–0.09, 0.22)
Supervisor support^a			–0.15 (–0.20, –0.10)			–0.17 (–0.24, –0.10)
High pace^a			0.11(0.06, 0.16)			0.14(0.05, 0.22)
Chemical hazard exposure^a			0.06(–0.001, 0.11)			0.09(0.005, 0.18)
Physiological hazard exposure^a			0.02(–0.04, 0.07)			0.18(0.10, 0.25)
Constant	2.46(2.40, 2.52)	2.05(1.16, 2.94)	2.79(2.02, 3.56)	2.57(2.46, 2.67)	1.64(.56, 2.73)	1.03(0.15, 1.90)
R ²	0.01	0.05	0.36	0.02	0.08	0.53
Observations	392	392	392	151	151	151

Standardised values.

In the Swedish case, the position of the confidence interval bounds suggests a likely positive association between agency contracts and injury odds, even though all models include 1. This reflects a situation of suppression, as the inclusion of PDR dimensions strengthens the effect of the agency contract. Estimated predicted probabilities based on Models 3 and 6 indicate a higher accident risk in the Swedish context in absolute terms: the average probability of an accident is 41% for agency workers and 26% for user firm employees in Sweden, compared to only 14% for agency workers and 6% for user firm employees in Poland (not shown in table). Rather than indicating more hazardous working conditions, this difference may reflect underreporting and/or a different conceptualisation of an accident in the Polish case, implying that the fewer reported injuries may be of higher severity.

Within the economic pressures block, the confidence intervals for all dimensions in both cases include one. However, assignment insecurity stands out due to its relative strength and the divergent direction of effects between the two cases. An increase of one standard deviation in assignment insecurity is associated with approximately 60% higher injury odds in the Polish case and 40% lower odds in the Swedish case. In the block of work disorganisation, individuals who had been consulted on workplace risks had nearly 70% lower odds of an injury in the Polish case, while exposure to chemical hazards and a high pace of work were associated with about 80% higher injury odds in the past year. In the Swedish case, the higher odds of an injury were linked to exposure to physiological and chemical hazards. The only notable interaction was found between contract type and pace of work in the Swedish case (p=.101); high perceived pace was associated with higher injury odds for user firm employees (17% vs 37%) but lower for agency workers (41% vs 34%) (not shown in table).

The OLS regression results in Table 4 show the associations for work-related exhaustion as the dependent variable. For reference in interpretation, a cut-off for high exhaustion levels using the OLBI has been set at 2.25 in previous literature (Schaufeli et al., 2001). The results indicate that agency workers experience lower exhaustion levels than user firm workers in both cases; however, the inclusion of low tenure and age reduces the effect of agency contracts in the Polish case and fully accounts for the contract effect in the Swedish case. The full models, which include elements of economic pressures and work disorganisation (Models 9 and 12), show a significant strengthening of the contract effect in the Polish case, suggesting that not all contract-related aspects are captured by the PDR dimensions. In the Swedish case, the regression coefficient for contract type remains unchanged, but the effect of low tenure decreases while the effect of age increases when PDR aspects are included. This suggests that, rather than contract type or tenure, work-related exhaustion is better explained by age and work-related characteristics. The final models, based on the PDR framework, significantly improve the explained variance in exhaustion, reaching 36% in the Polish case and 53% in the Swedish case.

Notably, all associations between aspects of economic pressures and work-related exhaustion are significant in the Polish case. In the Swedish context, assignment insecurity emerged as a significant predictor, with higher insecurity linked to higher exhaustion levels. Risk assessment and occupational health services had no effect on exhaustion in the Swedish case, whereas both had a protective effect in the Polish case. In both contexts, supervisory support was associated with lower exhaustion, while a high pace of work and exposure to chemical hazards predicted higher levels of reported exhaustion. Exposure to physiological hazards significantly increased exhaustion in the Swedish case but had no effect in the Polish case. To facilitate a qualitative cross-case comparison of effect patterns, regression outputs of the final logistic and OLS models were visualised in two combined graphs, shown in Figures 1 and 2.

Figure 1.

Representation of logistic model outputs.

Figure 2.

Representation of OLS model outputs.

Cross-case differences suggest a higher number of significant effects and larger effect sizes for elements of the PDR model for work-related exhaustion (Figure 2), particularly concerning contract type and economic pressures. This indicates that in the current cases, longer-term health outcomes, rather than injuries, are more strongly influenced by contextual differences. Moreover, national contexts appear to have had a greater impact on shaping divergent exposure to economic pressures than on work disorganisation.

Interaction effects revealed four significant interactions in the Polish case. Being consulted about workplace risks was associated with a reduction in exhaustion levels by 0.49 points for user firm employees, but it had no effect on agency workers (interaction effect, p=.03). An increase of two standard deviations in income insecurity was associated with a 0.38-point increase in exhaustion for agency workers, compared to only a 0.12-point increase for user firm employees (p=.09). A similar increase in job insecurity predicted a 0.46-point rise in exhaustion for agency workers and a 0.13-point increase for user firm employees (p=.056). Conversely, an increase in employment insecurity was linked to a 0.18-point decrease in exhaustion for user firm employees but a 0.09-point increase for agency workers (p=.09).

Discussion

Aligning with the prevalent pattern of findings from previous research (Håkansson et al., 2013), the article found that agency workers in both cases had higher odds of experiencing a work-related injury when compared to directly hired employees. Further analysis showed that these higher injury odds were not explained by agency employees’ younger age or shorter tenure, and the contract effect persisted in both cases even after controlling for aspects of work-related insecurity and work disorganisation. While it is not unexpected to find TAW jobs to be more hazardous in terms of injuries, the paper contextualises this finding, showing how local factors moderate accident exposure. Finding that aspects of economic pressures and work disorganisation were associated with injury odds independently of contract type serves as a reminder that all employees, regardless of their contract type, are at risk of workplace injury when exposed to these factors – something that is often forgotten when comparing precarious workers to those in ‘good’ standard employment.

Notably, injury odds were associated with the less-explored concept of assignment insecurity in terms of worry about being reassigned to another job against one’s will, rather than job or income insecurity, which is commonly linked to risk-taking and stress (Probst et al., 2018; Underhill and Quinlan, 2011). As is typical for a cross-sectional study that uses accidents as the dependent variable, an injury was measured as occurring prior (during the past year) to the current perception of insecurity. As such, the experience of an injury in the past year increased current assignment insecurity in the Polish case but decreased it in the Swedish case. Strauss-Raats (2019) identified a positive association between assignment insecurity and risk-taking in the Swedish context; however, that study did not link this finding to health outcomes. While the current article does not provide direct evidence on the role of institutional settings on workplace accidents, the findings suggest that the Swedish social security system and employer obligations for rehabilitation may act as a buffer in the association between an injury and perception of insecurity. In contrast, in the Polish case the lack of similar protections combined with the absence of employer obligations to support workers post-injury exposes workers to higher levels of insecurity.

Furthermore, agency workers in both cases reported lower levels of work-related exhaustion than user firm employees. The contract effect disappeared after accounting for PDR dimensions in the Swedish case but persisted in the Polish case. Qualitative data from the two cases suggest that the associations between contract type and exhaustion reflect differences in the organisational use of TAW. Differences in exhaustion levels along contract lines in the Polish context reflect a system of work organisation in which agency workers were segregated into simple tasks due to their short tenure and limited opportunities for upskilling. According to interview data, directing TAW into low-skill tasks increased task complexity, quality control demands and training obligations for user firm employees, which plausibly explains their higher exhaustion levels. Bryson (2013) has previously shown a positive association between higher work intensity in TAW user firms coupled with higher levels of anxiety among standard employees. The current article also highlights the increasing task complexity among user firm workforce as an OHS risk factor in the context of TAW use, particularly where there is workforce segmentation and TAW use as a flexibility buffer. Finding lower exhaustion levels among agency workers compared to user firm employees contrasts with previous studies (Dütsch, 2011; Wagenaar et al., 2012) and adds further ‘inconclusiveness’ to the associations between TAW and mental health (Hünefeld et al., 2020). However, through cross-case comparison and the incorporation of rich contextual data, this study enhances the understanding of the potential mechanisms behind health outcomes in TAW use, highlighting organisational integration as an important impact pathway.

Another example of work disorganisation as a significant driver of OHS outcomes was the notable lack of attention to OHS risks arising from task changes experienced by core employees. The interviewees reflected solely on the vulnerable position of agency workers but not the potential effects of the changing of jobs associated with TAW use, illustrating how risk pathways can be obscured in the context of increased organisational complexity. Furthermore, the article also demonstrates how the involvement of agency workers in the risk assessment process in the Polish context of workforce segmentation was inefficient as it did not impact injury exposure for agency workers, while user firm employees’ participation was associated with improved safety through lower injury odds. As suggested by Casey et al. (2017), increased organisational complexity is likely to disrupt system feedback mechanisms, with organisational segmentation emerging as a factor of complexity that may, in the long term, have negative consequences for the safety performance of the user organisation as a whole. Workforce externalisation was not identified as a risk in a risk assessment process in either case, with findings reinforcing the need for effective OHS management systems to consider organisational dimensions of OHS broadly. So far, both practice and research have paid relatively little attention to these aspects despite some early cautionary voices. Ward et al. (2001), for example, discussed employers’ tendency to rationalise TAW as an efficiency tool and to expect contract dualism while overlooking potential inter-group dynamics in shared workplaces, while Aronsson (1999) emphasised how workers’ precarious position weakened their role in effective OHS management as envisioned by Swedish legislation.

An area of OHS measures where both work disorganisation and economic pressures intersected was access to occupational health services. The Swedish case highlights unequal access to these services, with only one person among agency workers reporting a consultation. In the Swedish context, these consultations have rehabilitative rather than preventive purposes, confirmed by the positive association between consultation and accidents. The interviewees talked about barriers for agency employees in terms of physical access to the services, as the service facility was on-site for user firm employees but in a nearby town for the agency workforce. Another barrier that was mentioned was the fear of assignment change, with an interviewee describing a case where an agency employee in need of rehabilitation was reassigned rather than supported in returning to their work at the MNC. The vulnerability of agency workers to injury as well as the subsequent lack of support in return to work has been described in the Canadian context by Lippel and colleagues (2011). This vulnerability is partly explained by the regulatory system that incentivises outsourcing of hazardous jobs, as well as by the agencies’ need to preserve a relationship with the client and workers’ limited knowledge of their rights. The gap in access to occupational healthcare in the current case may indicate similar dynamics – reflecting either the economic pressure for the agency to reassign an injured worker rather than rehabilitate them to the same job or limited awareness on the side of workers to access the services. However, the current study mainly indicates a concerning pattern but did not collect sufficiently in-depth data on this dimension of OHS, leaving occupational healthcare in the context of TAW an area to explore in the future. Furthermore, our results show how organisational dynamics and interconnections between work and employment dimensions along the lines of economic pressures and work disorganisation, together with their effects on work-related health, remain areas in need of further research.

By showing the moderating effect of contract type on the effects of income, job and employment insecurity on work-related exhaustion, but only in cases where TAW was used on a short-term basis, segregated into the organisational periphery, and within a context that affords minimal protections to buffer insecurity effects, these findings support the notion that a stronger association exists between job insecurity and health in contexts with weaker social security protections (Kim et al., 2012; László et al., 2010). At the same time, the article shows that broader labour market insecurities compound problematic working conditions for precarious workers (Probst et al., 2018). It is notable, however, that the Swedish agency workers working on open-ended employment contracts perceived significant economic pressures. Even though this did not translate into higher levels of exhaustion, it highlights the risk of harm associated with the TAW contracts, even in the contexts typically described as ‘best case’ examples in the literature. The ‘objective’ security in an open-ended agency contract thus seems to hide significant subjectively perceived insecurities.

By demonstrating how TAW practices affect employees under both contract types, the current findings align with Lewchuk et al. (2008), who argue that rather than simply equating precarious contracts with detrimental health outcomes, it is essential to consider the variation in specific job characteristics and their interactions. Comparing different safety outcomes, the more similar cross-case effect patterns for injuries, compared to exhaustion, suggest that distinct contextual mechanisms shape OHS outcomes in the short and long term. This finding calls for caution when referring to ‘OHS’ in general terms.

Returning to the PDR model of workplace health (Quinlan et al., 2001) as a conceptual framework of this study, the elements of economic pressures and work disorganisation explained a considerable share of the variation in both work-related injuries and exhaustion. The model thus proves valuable for exploring OHS in organisational contexts where precarious forms of work, such as TAW, are utilised using survey tools. The impact of existing labour market regulations was evident in cross-case comparisons, with both liberal and restrictive regimes (Sartori, 2016) distinctly shaping job quality. More detrimental effects on job quality were observed in regimes that incentivise workforce segmentation and short-term TAW usage.

A notable overarching finding of the current study was that regulatory failure played a critical role in shaping OHS outcomes. The PDR model’s call for enhanced regulatory scrutiny of precarious work arrangements resonates with our findings, particularly in the Polish case, where regulatory gaps allowed for more pronounced segmentation between TAW and directly hired employees, often to the detriment of both groups. Overall, the findings emphasise that regulatory involvement needs to be proactive in addressing the risks inherent in both regulatory failures and work disorganisation. Mechanisms facilitating worker involvement are crucial to ensuring that all employees – regardless of their contract type – are protected within increasingly complex and segmented workplaces. This study also suggests that greater attention should be paid to how economic pressures and disorganised work environments interact to produce adverse health outcomes, particularly in contexts where performance-based pay and precarious work arrangements intersect.

Conclusion

This article has demonstrated that differences in job quality and OHS outcomes associated with the use of TAW are context-dependent, largely shaped by firm-level adaptations to existing regulatory contexts that impact on organisational practices related to TAW. While the current cases illustrate the commonly reported opportunity for employers to outsource risk (Håkansson et al., 2013; Underhill and Quinlan, 2011), they also highlight how workforce externalisation and segmentation can disrupt OHS management systems. Despite the relatively lower injury risk in the short term, the long-term health of user firm employees may be compromised. At the same time, agency workers remain vulnerable to economic pressures and the effects of work disorganisation, jeopardising their health both in the short and the long term. Therefore, the findings suggest that TAW use can potentially lead to a race to the bottom for working conditions, particularly in segmented labour markets. Thus, if aiming for harmonisation to ensure safe and healthy working conditions, current cases suggest that settings that facilitate the long-term strategic integration of TAW into the user firm are a better fit for the purpose and that institutional buffering of insecurity effects plays an important role in long-term health effects.

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This work was supported by the FP7-PEOPLE-2012-ITN project ‘Changing Employment’ (‘The changing nature of employment in Europe in the context of challenges, threats and opportunities for employees and employers’) under Grant no. 317321.

ORCID iD

Pille Strauss-Raats

Notes

Author biographies

Pille Strauss-Raats works at the Institute of Environmental Medicine, Unit of Occupational Medicine, Karolinska Institutet, Sweden. Her main research areas are precarious work, quality of working life, and employee wellbeing, as well as the effects of digitalisation and algorithmic management on working conditions.

Dragoș Adăscăliței is a research officer in the Employment Unit at the European Foundation for the Improvement of Living and Working Conditions (Eurofound). His research explores the future of work, with a particular focus on the impact of artificial intelligence on employment and working conditions, as well as emerging forms of work.

References

Ahlberg

(2008) Transnational Labour Regulation: A Case Study of Temporary Agency Work, Vol 60. Berlin: Peter Lang.

Arbetsmiljöverket (2013) Projektrapport: In- och uthyrning av arbetskraft. Available at: https://www.av.se/globalassets/filer/nyheter/in-och-uthyrning-av-arbetskraft–projektrapport2013-10-29.pdf (accessed 11 November 2020).

Aronsson

(1999) Contingent workers and health and safety. Work, Employment and Society 13(3): 439–459.

Berglund

Furåker

Vulkan

(2014) Is job insecurity compensated for by employment and income security? Economic and Industrial Democracy 35(1): 165–184.

Bohle

Quinlan

McNamara

et al. (2015) Health and well-being of older workers: Comparing their associations with effort–reward imbalance and pressure, disorganisation and regulatory failure. Work & Stress 29(2): 114–127.

Bryson

(2013) Do temporary agency workers affect workplace performance? Journal of Productivity Analysis 39(2): 131–138.

Byoung-Hoon

Frenkel

(2004) Divided workers: social relations between contract and regular workers in a Korean auto company. Work, Employment and Society 18(3): 507–530.

Casey

Griffin

Flatau Harrison

Neal

(2017) Safety climate and culture: Integrating psychological and systems perspectives. Journal of Occupational Health Psychology 22(3): 341–354.

Coe

Johns

Ward

(2009) Labour regulation, corporate strategies and market dynamics in the Swedish temporary staffing industry. European Urban and Regional Studies 16(1): 65–85.

10.

Davis-Blake

Broschak

George

(2003) Happy together? How using nonstandard workers affects exit, voice, and loyalty among standard employees. Academy of Management Journal 46(4): 475–485.

11.

Dellve

Skagert

Eklöf

(2008) The impact of systematic occupational health and safety management for occupational disorders and long-term work attendance. Social Science & Medicine 67(6): 965–970.

12.

Demerouti

Bakker

Vardakou

Kantas

(2003) The convergent validity of two burnout instruments: A multitrait-multimethod analysis. European Journal of Psychological Assessment 19(1): 12–23.

13.

Dütsch

(2011) Wie prekär ist Zeitarbeit? Eine Analyse mit dem Matching-Ansatz. Zeitschrift für Arbeitsmarktforschung 43(4): 299–318.

14.

EU-OHSA (2017) Estimating the costs of work-related accidents and ill-health: An analysis of European data sources. Luxembourg: Publications Office of the European Union.

15.

EU-OSHA (2022) Worker participation – Poland. Available at: https://oshwiki.osha.europa.eu/en/themes/worker-participation-poland (accessed 4 February 2025).

16.

Eurostat (2025) Temporary employment agency workers by sex, age and NACE Rev. 2 activity, lfsa_qoe_4a6r2. Available at: https://doi.org/10.2908/LFSA_QOE_4A6R2 (accessed 31.03.2025)

17.

Ferrie

Westerlund

Virtanen

et al. (2008) Flexible labour markets and employee health. SJWEH Supplements 6: 98–110.

18.

George

Bennett

(2005). Case Studies and Theory Development in the Social Sciences. Cambridge, MA: MIT Press.

19.

Greenland

Senn

Rothman

et al. (2016) Statistical tests, P values, confidence intervals and power: a guide to misinterpretations. European Journal of Epidemiology 31(4): 337–350.

20.

Håkansson

Isidorsson

(2016) Between two stools: Occupational injuries and risk factors for temporary agency workers. International Journal of Workplace Health Management 9(3): 340–359.

21.

Håkansson

Isidorsson

Kantelius

(2012) Temporary agency work as a means of achieving flexicurity? Nordic Journal of Working Life Studies 2(4): 153–169.

22.

Håkansson

Isidorsson

Strauss-Raats

(2013) Kunskapssammanställning: arbetsmiljö för hyresarbetskraft: inhyrdas fysiska och psykosociala arbetsmiljö. Stockholm: Arbetsmiljöverket.

23.

Hintikka

(2011) Accidents at work during temporary agency work in Finland: Comparisons between certain major industries and other industries. Safety Science 49(3): 473–483.

24.

Holst

Nachtwey

Dörre

(2010) The strategic use of temporary agency work: Functional change of a non-standard form of employment. International Journal of Action Research 6(1): 108–138.

25.

Hünefeld

Gerstenberg

Hüffmeier

(2020) Job satisfaction and mental health of temporary agency workers in Europe: A systematic review and research agenda. Work & Stress 34(1): 82–110.

26.

Jann

(2014) Plotting regression coefficients and other estimates. The Stata Journal 14(4): 708–737.

27.

Kim

Muntaner

Shahidi

et al. (2012) Welfare states, flexible employment, and health: A critical review. Health Policy 104(2): 99–127.

28.

László

Pikhart

Kopp

et al. (2010) Job insecurity and health: A study of 16 European countries. Social Science & Medicine 70(6): 867–874.

29.

Leka

Jain

Zwetsloot

Cox

(2010) Commentary: Policy-level interventions and work-related risk management in the European Union. Work & Stress 24(3): 298–307.

30.

Leveson

(2015) A systems approach to risk management through leading safety indicators. Reliability Engineering & System Safety 136: 17–34.

31.

Lewchuk

Clarke

De Wolff

(2008) Working without commitments: Precarious employment and health. Work, Employment and Society 22(3): 387–406.

32.

Little

Rubin

(2019, Statistical Analysis with Missing Data, Vol. 793. Chichester: John Wiley & Sons.

33.

Lippel

Ellen

Ron

et al. (2011) Legal protections governing the occupational safety and health and workers’ compensation of temporary employment agency workers in Canada: Reflections on regulatory effectiveness. Policy and Practice in Health and Safety 9(2): 69–90.

34.

Marino

Bernaciak

Mrozowicki

Pulignano

(2019) Unions for whom? Union democracy and precarious workers in Poland and Italy. Economic and Industrial Democracy 40(1): 111–131.

35.

Mize

(2019) Best practices for estimating, interpreting, and presenting nonlinear interaction effects. Sociological Science 6: 81–117.

36.

Mrozowicki

Maciejewska

(2016) The rise of the dual labour market: Fighting precarious employment in the new member states through industrial relations (PRECARIR) Country report: Poland, no. 13. Bratislava: Central European Labour Studies Institute (CELSI).

37.

Muñoz de Bustillo

Fernández-Macías

Antón

Esteve

(2009) Indicators of job quality in the European Union. Brussels: European Parliament’s Committee on Employment and Social Affairs.

38.

Nielsen

Taris

Cox

(2010) The future of organisational interventions: Addressing the challenges of today’s organizations. Work & Stress 24(3): 219–233.

39.

Pejtersen

Kristensen

Borg

Bjorner

(2010) The second version of the Copenhagen Psychosocial Questionnaire. Scandinavian Journal of Public Health 38(3_suppl): 8–24.

40.

Probst

Brubaker

(2001) The effects of job insecurity on employee safety outcomes: Cross-sectional and longitudinal explorations. Journal of Occupational Health Psychology 6(2): 139–149.

41.

Probst

Petitta

Barbaranelli

Lavaysse

(2018) Moderating effects of contingent work on the relationship between job insecurity and employee safety. Safety Science 106: 285–293.

42.

Quinlan

(2015) The effects of non-standard forms of employment on worker health and safety. Geneva: International Labour Office.

43.

Quinlan

Bohle

(2009) Overstretched and unreciprocated commitment: Reviewing research on the occupational health and safety effects of downsizing and job insecurity. International Journal of Health Services 39(1): 1–44.

44.

Quinlan

Mayhew

Bohle

(2001) The global expansion of precarious employment, work disorganisation, and consequences for occupational health: A review of recent research. International Journal of Health Services 31(2): 335–414.

45.

Sartori

(2016) Temporary agency work in Europe: Degree of convergence following Directive 2008/104/EU. European Labour Law Journal 7(1): 109–125.

46.

Schaufeli

Bakker

Hoogduin

et al. (2001) On the clinical validity of the Maslach Burnout Inventory and the Burnout Measure. Psychology & Health 16(5): 565–582.

47.

Seawright

(2016) Multi-Method Social Science: Combining Qualitative and Quantitative Tools. Cambridge: Cambridge University Press.

48.

Strauss-Raats

(2019) Temporary safety: Regulating working conditions in temporary agency work. Safety Science 112: 213–222.

49.

Sverke

Hellgren

Näswall

(2002) No security: A meta-analysis and review of job insecurity and its consequences. Journal of Occupational Health Psychology 7(3): 242–264.

50.

Sverke

Låstad

Hellgren

et al. (2019) A meta-analysis of job insecurity and employee performance: Testing temporal aspects, rating source, welfare regime, and union density as moderators. International Journal of Environmental Research and Public Health 16(14): 2536.

51.

Szumilas

(2010) Explaining odds ratios. Journal of the Canadian Academy of Child and Adolescent Psychiatry 19(3): 227–229.

52.

Thörnquist

(2008) The Swedish occupational safety and health model lost in transition? In: OSH & Developments. Stockholm: The Swedish Association for Occupational and Environmental Health & Development, pp. 43–62.

53.

Underhill

Quinlan

(2011) How precarious employment affects health and safety at work: The case of temporary agency workers. Relations Industrielles/Industrial Relations 66(3): 397–421.

54.

Vosko

(2011) Precarious employment and the problem of SER-centrism in regulating for decent work. In: Lee

McCann

(eds) Regulating for Decent Work. London: Palgrave Macmillan, pp. 58–90.

55.

Wagenaar

Kompier

Houtman

et al. (2012) Can labour contract differences in health and work-related attitudes be explained by quality of working life and job insecurity? International Archives of Occupational and Environmental Health 85(7): 763–773.

56.

Ward

Grimshaw

Rubery

Beynon

(2001) Dilemmas in the management of temporary work agency staff. Human Resource Management Journal 11(4): 3–21.

57.

Wasserstein

Lazar

(2016) The ASA statement on p-values: context, process, and purpose. The American Statistician 70(2): 129–133.

58.

Wilkin

de Jong

Rubino

(2018) Teaming up with temps: the impact of temporary workers on team social networks and effectiveness. European Journal of Work and Organizational Psychology 27(2): 204–218.

59.

Williams

(2012) Using the margins command to estimate and interpret adjusted predictions and marginal effects. The Stata Journal 12(2): 308–331.

60.

Williams

(2015) Missing data part 1: Overview, traditional methods. University of Notre Dame. Available at: https://www3.nd.edu/~rwilliam/stats3/MD01.pdf (accessed 11 November 2020).

The hidden costs of flexibility: A comparative study of OHS outcomes in using temporary agency work

Abstract

Keywords

Introduction

Contextual drivers in OHS: PDR model of workplace health

Economic pressures

Work disorganisation

Regulatory failure

Contrasting regulatory contexts: Sweden and Poland

Materials and methods

Research design, case organisation and participants

Survey variables

Empirical approach

Results

Qualitative data: Organisational practices of TAW use and OHS

Poland: TAW as a flexibility buffer

Sweden: Strategic use of TAW

Survey results

Discussion

Conclusion

Footnotes

Declaration of conflicting interests

Funding

ORCID iD

Notes

Author biographies

References