Impact of different work organizational models on gender differences in exposure to psychosocial and ergonomic hazards at work and in mental and physical health

Although the relationship between quality of work and health has been investigated for several decades, the issue of women’s well-being in the workplace is still an under-researched area (Connerley and Wu 2016). Female work, especially manual one, is often characterized by monotonous, repetitive actions, with static effort and multiple simultaneous responsibilities, which potentially threaten both women’s physical and mental health; work areas, tools, and pace derive from a work organization and an equipment endowment created for a male population and may not be suitable for women, who have a different anthropometric structure (Bond et al. 2004; Messing 1999, 2017).

Different epidemiological studies have found women at higher risk of developing some work-related diseases, especially musculoskeletal (de Zwart et al. 2001; Hagberg and Wegman 1987; Park et al. 2017; Roquelaure et al. 2006) and mental disorders (Beauregard et al. 2018; Bilodeau et al. 2020; Blehar 2006; Kuehner 2003; Roxburgh 1996; Wege et al. 2018). Although the mechanisms underlying such excess risks are still unclear, it has been proposed that they would stem, at least in part, from differential exposure to physical and psychosocial factors at work between men and women (Krieger 2003; Quinn and Smith 2018).

In general, from the literature, women appear more exposed than men to the hazards currently most diffuse among workers, i.e., ergonomic and psychosocial ones, including awkward and tiring postures (Eng et al. 2011; Hooftman et al. 2005; Nordander et al. 1999), hand/arm repetition (Sterud 2014), standing and bending (Sterud 2014), walking long time (Bauer et al. 2009), low job control (d’Errico et al. 2011; Hooftman et al. 2005; Josephson et al. 1999; Messing et al. 2009), high demand (Sterud 2014), high job strain (d’Errico et al. 2011; Ibrahim et al. 2001; Karlqvist et al. 2002), repetitive work (Eng et al. 2011; Messing et al. 2009; Nordander et al. 1999; Strazdins and Bammer 2004), effort–reward imbalance (Johannessen and Sterud 2017; Sterud 2014), and sexual harassment (Das 2009; Messing et al. 2009; Sterud 2014).

However, the results on exposure to adverse work factors appear inconsistent among studies, as shown by a systematic review on the subject, which found a higher prevalence of exposure in women only for job insecurity, low job control, and worse contractual working conditions (Campos-Serna et al. 2013).

The higher exposure of women to workplace ergonomic and psychosocial factors has been interpreted mainly as attributable to women’s work segregation, i.e., the selective employment of women in certain economic sectors and in lower status jobs, in terms of professional position and responsibilities. Two types of work segregation are commonly distinguished: horizontal and vertical. The horizontal segregation means that women are generally employed in sectors and jobs different from those of men. The economic sectors with the highest concentration of female workers are health care, trading, education, tourism, services, and domestic work, whereas men are mainly employed in construction, manufacturing, transportation, agriculture, and finance. Few sectors can be actually defined as mixed, such as the public sector and some branches of manufacturing, like textiles and garment or food production, with women more often employed in small companies. The vertical segregation means that women are less represented in executive or more remunerated positions, with hierarchies at work clearly dominated by men, who generally have higher wages, more secure jobs, and more career perspectives (Blau and Kahn 2007; Fagan and Burchell 2002). Vertical segregation is mainly driven by two mechanisms which limit the possibility for women to access higher positions. The first one has been named “glass ceilings”, intending that for female workers to be promoted at top level positions is very difficult because of invisible barriers created by men’s power and attitudes. The second is that of “sticky floors”, meaning that women have more difficulties than men in leaving entry-level job for higher positions (Pyle and Bond 1997). In this framework, it is pointed out that even if they are employed within the same job title, women and men usually perform different tasks and activities (Messing and Mager Stellman 2006), which may expose them to different types of occupational hazards, or to the same ones, but with different exposure intensity and frequency (Messing et al. 1994).

However, the gender segregation approach tends to emphasize the issue of hierarchy, power, and position in it, neglecting other organizational aspects relevant to the gender perspective (Quinn and Smith 2018), such as behavioural expectations concerning languages, emotions, competition, performance, collaboration, assertiveness, rationality, control, and autonomy. Other perspectives deal with these aspects and argue that organizations are gendered, as they have been “designed by men for men”, that is, they are largely defined by male practices (Acker 2012; Benschop and Verloo 2016; Burke 2014; Kelan 2018; Lewis et al. 2017; Rumens 2017; Wahl 2014). Even when managers are women, organizations tend to remain gendered in favour of men’s practices. This happens, because the managerial practices are powerfully and historically associated with a masculinist way of being and behaving, and all managers are required to conform to them (Whitehead 2014). This mechanism can explain the slowing pace of change toward gender equality at work (Benschop and Van den Brink 2014).

In this context, besides gender segregation, also employment in firms with different models of work organization may influence women’s well-being and exposure to workplace hazards, compared to their male counterparts. Even if gender equality is far to be achieved, the different types of work organization can create different social and cultural environments, and therefore bring about different well-being conditions for women (and men).

The present study aims at contributing to the debate on the role of different types of work organization models as determinants of exposure to physical and psychosocial factors in the workplace, and of workers’ health and well-being in the European Union, with a strong focus on gender differences. More specifically, objective of the study is to assess whether different models of work organization display gender differences in self-reported exposure to psychosocial and ergonomic factors at work, and in self-reported mental and musculoskeletal health. Our research builds on the work by Lorenz and Valeyre (2005), who used several work characteristics collected in the European Working Conditions Survey (EWCS) 2000 to identify clusters of workers employed in work organizations belonging to the different models. The advantage of adopting an approach based on work organization models rather than on gender-based segregation to assess gender differences in workplaces, is that: (1) is closer to industrial and economic discourses oriented to enhancement of productivity in different types of work organization; and (2) there are growing doubts that desegregation could always benefit women (Messing 2017).

Another aim of the study was to assess whether among European workers gender differences in exposure to work factors and in work-related health changed between 2010 and 2015, also using EWCS data.

Socio-demographic characteristics of the study population in 2010 and 2015, divided by gender and type of work organization, are presented in Tables 1 and 2, respectively. The HCPC on 2010 data identified three clusters as the best number for clustering subjects, as we did not find empirical evidence for the fourth one, the traditional model. In contrast, using 2015 data, four clusters were identified, corresponding to the four organizational models, as expected.

One cluster was composed of 3595 subjects in 2010 and 2777 in 2015, and was characterized by a low degree of horizontal constraints, normative constraints, automatic constraints, repetitiveness, monotony, and quality norms, but lower team work and job rotation (Tables 1 and 2); based on such features, it was interpreted as “reflexive production”.

A second cluster, interpreted as “lean production”, was composed of 4551 subjects in 2010 and 3619 in 2015, and displayed high levels of team work, job rotation, time autonomy, problem-solving, and complexity, although also reporting relatively high levels of several types of constraints indicating exposure to high work pressure, such as presence of norms on quantity and quality of the production performed, tight control by supervisors, and dependency on the work pace of machines and colleagues (Tables 1 and 2).

A third cluster was composed of 1562 subjects in 2010 and 2460 in 2015, and had high levels of horizontal constraints, normative constraints, automatic constraints, repetitiveness, monotony, and low levels of time autonomy, methods autonomy, learning, and problem-solving (Tables 1 and 2); based on these characteristics, this cluster was interpreted as “Tayloristic production”.

The fourth cluster, found only in 2015 data, was also characterized by low levels of time autonomy, methods autonomy, learning, and problem-solving, but repetitiveness and monotony were less prevalent than in Tayloristic production, while levels of horizontal and normative constraints were very low, similar to those of the reflexive production model.

Examining diffusion of the organizational models by European region, in EWCS 2015, lean production was found to be the most common model in the Anglo-Saxon (43.9%) and the Scandinavian regions (41.7%), areas where the traditional model was the least common (10.7% and 10.0%, respectively), while reflexive production was highest in the Scandinavian and Continental regions (36.4% and 34.0%, respectively), and Tayloristic production in the Eastern, Southern, and non-EU regions (26.8%, 27.4% and 26.8%, respectively). For traditional production, the highest prevalence was seen in the Southern and Eastern regions (18.7% and 16.8%, respectively).

In Table 3, prevalences by gender, as well as female-to-male prevalence ratios (PRs) of exposure to psychosocial and ergonomic factors at work are presented for each work organization model and year of the survey. Except for exposure to high job strain in 2015, for all other hazards, the prevalence of exposure was lower among both male and female workers employed in firms adopting the reflexive production model, than among those employed in companies with other organizational models.

In both surveys, women had a higher likelihood of exposure than males to unfavourable work factors in all types of work organization, except for carrying or moving heavy loads. Only for job strain and moving/carrying heavy loads results consistently showed across surveys an exposure profile more favourable to women in the reflexive production than in the lean and Tayloristic production.

In detail, for high job strain in 2010, the female/male PRs were significantly higher in the lean production (PR = 1.26, 1.14–1.40) and in the Tayloristic production (PR = 1.16, 1.10–1.22), compared to reflexive production (1.06–0.94–1.20) (p value for interaction = 0.07 and 0.03, respectively), whereas in 2015, only the PR for the lean production was significantly higher (PR = 1.43, 1.19–1.73 vs. PR = 1.14, 0.95–1.3; p value for interaction = 0.07). For ERI, in 2010, the female/male PR of the reflexive production (PR = 1.03, 0.91–1.6) was significantly lower than that of the other two models (lean: PR = 1.18, 1.08–1.29; Tayloristic: PR = 1.25, 1.15–1.37) (p value for interaction = 0.10 and 0.003, respectively), whereas in 2015, no significant differences were observed among the four models.

Regarding ergonomic hazards, in 2010, the female/male PR for awkward postures observed in reflexive production was significantly lower than that in Tayloristic production (PR = 1.00, 0.87–1.16 vs. PR = 1.24, 1.14–1.36; p value for interaction = 0.003), whereas no significant difference was observed in 2015. Also, in both surveys, a significantly lower female/male PR for carrying/moving heavy loads was found in reflexive production compared to lean production (PR = 0.48, 0.38–0.59 vs. PR = 0.62, 0.53–0.74 in 2010; PR = 0.73, 0.57–0.95 vs. PR = 0.99, 0.82–1.19 in 2015), although in 2015, it was higher than that observed in the traditional model (PR = 0.44, 0.33–0.59). Finally, no significant differences were observed for repetitive movements across the three models in 2010, whereas in 2015, gender PR were highest in reflexive production, and significantly higher than in Tayloristic production (PR = 1.56, 1.40–1.73 vs. PR = 1.21, 1.14–1.29; p value for interaction = 0.03).

Regarding health, both low mental well-being and musculoskeletal pain were more prevalent in females than males (Table 4). For low mental well-being, in 2010, gender PRs did not show any significant difference across the different organizational models, whereas in 2015, the PR was highest in reflexive production (PR = 1.49, 1.22–1.81), and significantly higher than that in traditional production (PR = 1.13, 0.89–1.42). For musculoskeletal pain, no significant differences in gender PRs between the different organizational models were observed in the 2015 survey, whereas in 2010 the female/male PR of backache in reflexive production was significantly lower than that in Tayloristic production (PR = 1.06, 0.98–1.14 vs. PR = 1.18, 1.08–1.30; p value for interaction = 0.03), while that of upper limb musculoskeletal pain was significantly lower than that in lean production (PR = 1.16, 1.06–1.25 vs. PR = 1.24, 1.16–1.33; p value for interaction = 0.05).

The comparison between 2010 and 2015 overall samples showed similar prevalences of exposure to work factors and of health outcomes in both men and women, except for low mental well-being, whose prevalence decreased in 2015 by almost 5% in men and 7% in women (Table 5). No significant difference in gender PRs for any work factor or health condition was found between the two surveys.

In general, female/male PRs of exposure to work factors and of health outcomes were stronger in Continental, Eastern, Scandinavian, and non-EU countries, while lower in Anglo-Saxon and Southern countries, with significant differences among regions for all the variables examined, except for low mental health and musculoskeletal pain in the upper limb (Supplementary Table 3). In contrast, Gender PRs did not differ significantly between EU and non-EU countries, except for carrying/moving heavy loads, to which women were less exposed in non-EU countries (data not shown).

As expected, in European workplaces, women were more likely than men to be exposed to the psychosocial and physical hazards examined, except for carrying or moving heavy loads, and were more likely to report low mental well-being and musculoskeletal pain.

Regarding differences in exposure to work hazards between men and women across the different organizational models, we found in both 2010 and 2015 surveys that the female/male prevalence ratio of exposure to work stress, defined according to the job demand-control model (Karasek 1979), was significantly lower in workplaces characterized by the reflexive production model, compared to those employing the lean production. In contrast, results for ERI were discordant between the two surveys, with significantly or marginally significantly higher gender PRs in lean and Tayloristic production, compared to reflexive production, and no differences in 2015. With respect to ergonomic hazards, smaller differences were observed by gender, although in both surveys, the female/male PRs of exposure to carrying/moving heavy loads were significantly lower in reflexive production than in lean production. Concerning health, in 2010, gender differences in musculoskeletal pain were more favourable to women in reflexive production, compared to the other two models, even if with some differences for back and upper limb pain, whereas no differences were present in 2015. For low mental well-being, no differences in the gender PRs were present across the three models in 2010, although in 2015, the PR in reflexive production was higher than that of all other models, and significantly different from that of traditional production.

In summary, we found consistent evidence across the two surveys only for a lower female/male ratio of exposure to job strain and to heavy loads in reflexive production, compared to lean production. It seems unlikely that differences in the results observed between the two surveys could derive from actual changes in working conditions occurred during this period, given the relatively short time elapsed, but rather to chance, also considering that interactions between gender and organizational model in 2010 for other work exposures and for health conditions were mostly of borderline significance. However, gender PRs in the overall sample were quite consistent across surveys, suggesting that differences in the number of organizational models identified in 2010 and 2015 may have contributed to limit comparability of the results by type of organization between the two surveys.

A possible interpretation of the gender differences in exposure to job strain observed between reflexive and lean production is that the reflexive production organization is less gendered. Indeed, some scholars have mentioned the learning organization as a type of organization more conducive to women, as it is able to construct an environment of learning, which gives value to the individuals and their competences and experiences (Cram et al. 2016; Johansson and Abrahamsson 2018; Luciano 2008; Raaijmakers et al. 2018). This type of organizational context would be more open to value diversity and therefore to women’ culture and attitudes. However, learning new things and problem-solving in our analysis were both more diffuse in the lean production model. Concerning workplace learning, it is possible that the employees do not recognize ways of more informal learning occurring in their daily work (Evans 2002). The framing of the question in the questionnaire—“learning new things”—might be interpreted by the responders as referring to explicit learning activities. In the lean production organization, learning could be more evident, thanks to the constant request and effort to improve the work. Workplaces organized in accordance with the reflexive production model may be characterized by a subtle way of learning (Billett 2004), less recognizable by the workers but however present, linked to a more interdisciplinary, integrated and interactive work (Lundvall et al. 2011). These features may create a workplace environment accustomed to diversity and different points of view, and therefore more favourable to female employees. According to some authors, learning in this type of work organization is an experienced-based learning, emerging from “doing, using and interacting”, characterized by strong elements of tacit knowledge. This happens thanks to multidisciplinary workgroups, integration of functions, and closer interaction with customers (Lundvall et al. 2011).

Looking closer at the differences between reflexive and lean production highlights some more aspects relevant to interpret the gender gap in exposure to job strain. The only features which seem to explain the lower female/male PRs of exposure to job strain in the reflexive production model are related to the higher monotony, higher repetitiveness, and higher exposure to different types of constraints, all characteristics more diffused in lean production. In particular, it seems likely that more strict quantitative production and quality norms in lean production, together with direct control by supervisors, would increase work intensity and effort requested to workers, an effect already pointed out in the literature critical toward the lean production. Different studies on the subject found that this type of work organization brings about intensified work pace and demands, and job strain (Arezes et al. 2015; Bouville and Alis 2014; Landsbergis et al. 1999; Oudhuis and Tengblad 2020; Stewart et al. 2016). Yet, a growing literature points out the role of management practices in the way lean production is implemented, which may to limit work intensification and workers’ strain (Bocquet et al. 2019; Koukoulaki 2014; Longoni et al. 2013; Neirotti 2020; Stimec and Grima 2019), although this might not work in certain sectors (Ogbonnaya et al. 2017). Such a work intensification could be experienced especially by female workers, who are more segregated in lower hierarchical positions, and this in turn would raise their exposure to work stress and ergonomic hazards. Some scholars have already noted that the introduction of organizational forms of lean production does not favour women, nor it affects gender segregation (Abrahamsson 2014; Losonci et al. 2011; Zanoni 2011). Although diverse applications of lean production exist, depending on cultural and institutional contexts, with different mixes of elements of the ideal-type of lean production, in general, the involvement of workers appears as problematic; the studies of Babson (1995), Rinehart and colleagues (1997), and Appelbaum and Batt (1993) have found that work in lean production is actually organized in a way that often is accompanied by job security reduction, lack of promotion, and weak representation of workers’ interests. This problematic aspect of the lean production model seems to persist: recent works still discuss the need for improvement of the workers’ participative processes, their autonomy, and their learning (Lantz et al. 2015; Stimec and Grima 2019), which are lacking where the lean production principles are misapplied (Neirotti 2020), a situation which appears to be frequent (Arezes et al. 2015). The finding in our study that both men and women employed in companies belonging to reflexive production reported a lower prevalence of exposure to psychosocial and physical hazards, compared to the other types of work organization, gives support to the theory that such a model is characterized by working conditions more acceptable to workers of both genders.

The lesser job constraints experienced in the reflexive production model and the much lower frequency of teamwork could be the main features which favour the lower female/male prevalence ratio of work stress in this organizational model. One interpretation could be that these features allow women to enjoy more freedom and higher control on their jobs, and that they can reduce the pressure of a male-gendered organization with its dominant male-culture and practices. This would decrease women’s exposure to psychological demands and increase job control, with a consequent reduction in their level of job strain, compared to other organizational models. In particular, the lower frequency of teamwork could give women even more chance to avoid stress and lack of recognition. Williams and colleagues (2012) note that teamwork, very often supervised by male staff, tends to obscure individual contribution and put more stress on women to promote themselves and to receive credit from their supervisors and peers. As much literature has demonstrated, women find difficulties in receiving non-paternalist support in their carrier and “are given disproportionately less credit than men for the success they achieve when they work on teams in male-dominated environments …” (ibidem, 557). It is possible to speculate that, given that too often the organizational processes are still governed by gender prejudices and stereotypes, in flatter organization teamwork can play in women’s disfavour. Feminist theories argue that women need freedom in a male-dominate environment to express themselves in autonomy with respect to gender roles socially constructed (Bertell 2016; Dini and Tarantino 2014; Youngblood Jackson 2013).

Among strengths of this study, it employed a large representative sample of the European working population, which on one hand provided the study with substantial statistical power, and on the other hand permits to generalize the findings to private employees living in the countries included in the survey. Although information on the work environment was self-reported, questions used in EWCS surveys to assess exposure to psychosocial and physical factors at work have been extensively validated (Wikman 1991), and in the last decades, many studies have used EWCS data for conducting occupational and social epidemiological research.

Nonetheless, the self-reported nature of information on working conditions and health does not allow to exclude that the higher prevalences observed among women may be attributable to an overestimation of exposure and health conditions in women, compared to men. However, as commented in the introduction, several studies have found, consistently with our results, higher exposure to most physical and psychosocial factors at work among women. Also for mental and physical health, women show in the literature a consistently higher likelihood of depressive symptoms (Salk et al. 2017) and musculoskeletal pain (Andorsen et al. 2017; de Zwart et al. 2001), compared to men. Another limitation is the lack of information on domestic workload, in terms of household duties and child care, as the greater family burden sustained by women (Anxo et al. 2011) may have concurred in determining the higher prevalences observed among women than men, in particular for health outcomes (Bilodeau et al. 2020; Beauregard et al. 2018).

Regarding the cross-sectional design of the study, as we compared exposure to work factors and health outcomes between men and women, overall and by organizational model, without investigating associations between exposure and health outcomes, the lack of temporality characteristic of this type of study is not expected to have biased in an important way the results. However, it seems difficult to exclude that women and men in the samples analyzed have been subjected to a different degree of selection in and out of the workforce.

In conclusion, a few gender differences were consistently observed in exposure to adverse physical and psychosocial factors at work among the different organizational models examined. However, the lower female/male ratio of exposure to job strain observed in both surveys in reflexive production, compared to lean production, seems to indicate that this work organizational model may be favourable to women, possibly because of a more limited amount of team work and a lower degree of hierarchical constraints characteristic of this type of organization.

This study suggests that the adoption of a work organization point of view allows to elaborate further on gender differences in well-being and health in workplaces, compared to the studies considering only type of industry, job title, tasks, and activities. The work organization concept permits to make sense of bundles of working conditions and open the discourse up to include the gendered dimension of work organization. Compared to the gender-based segregation approach, it allows to consider that even in a male-dominated work environment, the work organization could present features which grant more freedom to women. This may be the case of the reflexive production, where teamwork is less adopted, although, based on our results, we can only speculate that less teamwork plays in favour of women’s health. Our results indicate that female well-being in the workplace needs to be investigated further, drawing attention to specific gender issues. The literature on gendered organization theory seems one of the perspectives more promising in providing advancement to the field.