Opportunities for recovery at work and excellent work ability - a cross-sectional population study among young workers

This is a cross-sectional population study using the Work Environment Survey by Statistics Sweden on behalf of the Swedish Work Environment Authority. This survey is based on the Labour Force Survey and includes additional questions asked in a telephone interview and a follow-up postal questionnaire. The purpose of the survey is to describe the physical and psychosocial work environment of the employed population. The employed population was defined as those aged 16–64 years who worked at least one hour during the measurement week. The Work Environment Survey has been conducted every second year since 1989. For each of the years 2009, 2011, and 2013 the randomized selected sample was about 10 000–16 000, of whom approximately 8100– 12 400 answered the telephone interview and about 4800–7800 also answered the postal questionnaire. The telephone interview asks participants background questions on employment, work strain, leadership, and work ability. A week later the postal questionnaire asks about 121 specific occupational and health items. The development and validation of the method is described in a report from Statistics Sweden [15].

In 2009–2013 about 19 000 individuals responded to the full telephone interview. Of these 4949 were young workers, 18–29 years of age. The study sample consists of 2351 young workers who responded to the telephone interview, the postal questionnaire, and the WAS question in the telephone interview (Fig. 1). There was no overlap of individuals in the three years chosen for the study. Inclusion criteria for the study sample were age range 18–29 years, and having answered the single WAS question. For the subsample of individuals with high work demands, N = 788, the criterion was to have reported having either a job that was generally physically strenuous to a large extent or a high workload with far too much to do to a large extent, or both. The selection of this subsample was chosen as women and men with high work demands could have an increased need of recovery opportunities at work [14].

Individual characteristics drawn from the survey were used to describe the sample: sex, age, education, years in present occupation, trade, work ability, and health. The variable educational level was dichotomized into compulsory school or high school versus post-gymnasium, college, or university. Three health variables were identified: pain in at least one location including upper back or neck, lower back, shoulder or arms, wrists or hands, and hips, legs, knees or feet; feeling very tired after work; and sleep difficulties due to thoughts of work. These health variables were ranked by frequency: every day, a few days per week, a day per week, a few days per month, seldom or not at all. The three first alternatives implied impaired health.

High work demand was seen as a possible confounder for the analyses of the total sample to answer the first research question. This variable was, however, also used for creating a subsample, as described in the study sample section, to answer the third research question. For the second research question, two possible confounders were considered: high work demands and educational level. High work demands can influence work ability [21] and have a negative association with recovery opportunities at work [14] and was defined in terms of physically strenuous work to a large extent and/or high workload with far too much to do to a large extent. Poor basic education has been shown to relate to poor work ability among young adults [13] and was therefore considered as a confounder.

SAS version 7.11 (SAS Institute, Cary, NC, USA) was used for the analyses in this study. In all analyses the data were stratified for women and men [22]. A subsample of those with high work demands was also analyzed. Descriptive data were obtained through frequency analyses.

The associations between recovery opportunities at work and excellent work ability was analyzed with logistic regressions (proc genmod in SAS). The effect measures retrieved from these regressions were prevalence ratios (PRs) instead of the commonly used odd ratios (ORs). This is due to the outcome being common and hence ORs cannot be interpreted as approximations of PRs. From the logistic regressions PRs were calculated as the effect of the exposure around the prevalence of 50 %. In the presented PRs the group with the lower degree of recovery opportunities at work was used as the reference category. The 95 % confidence intervals (CIs) for PRs were calculated using the delta method [23].

A guideline of applied logistic regression was followed [24] to build models for testing associations between recovery opportunities at work and excellent work ability. Co-linearity was checked by examining pairwise correlations by cross tables considering > 80 % in a diagonal or any cells with no answers. This was performed between all exposure variables, between all exposure variables and the possible confounders of high work demands and educational level, and between the two confounders. “Being involved in planning the work” showed co-linearity with the variable “Possibility of deciding when to perform a work task”. “Being involved in planning the work” was therefore excluded from further analyses due to more missing compared to “Possibility of deciding when to perform a work task”. Univariate logistic regression analyses were then performed to select variables for the multiple regression with the criteria of p < 0.25.

The steps in the guideline for creating multiple models set criteria for variable inclusion at p < 0.25, changes in parameter estimates for exposure variables at < 15 %, and likelihood ratio tests between nested models. If an explanatory variable had p ≥ 0.25, but changed parameter estimates for other explanatory variables when the models were considered with and without it, it was included as a confounder. Possible interactions between exposure factors were also tested.

This study was approved by the Regional Ethical Review Board at the University of Gothenburg, Sweden (Reg.no. 221–15).

The proportion of women was 55 % in the study sample and 56 % in the subsample (Table 1). High work demands were reported by 34 % of the women (N = 439) and 33 % of the men (N = 349). Self-reported work ability was high for women and men (9.2) overall and in the group with high work demands (9.0–9.1). The young workers, especially women, generally had few years of work experience. The men seemed to have lower educational levels than the women, especially among those with high work demands.

The young women and men tended to work in different trades (not shown in the tables). The largest group of women, both in the whole sample and in the subsample with high work demands, was found in the service, health care, and retail sector. Men were most often employed in construction and manufacturing, and then in the service, health care, and retail sector, though men with high work demands worked mainly in construction and manufacturing.

Women with high work demands reported the poorest health (pain, fatigue, and sleep difficulties), although pain and fatigue were common in the whole sample (Table 1).

Recovery opportunities at work were reported at various levels, as about half of the study sample experienced some recovery at work. Those with high work demands, especially women, seemed to report fewer of several of the investigated recovery variables than the sample as whole. Possibilities of deciding one’s own working hours and to have varied work appeared to be quite similar for women and men in general, but at least the first variable seemed to be less common among those with high work demands.

Univariate associations between recovery opportunities at work and excellent work ability for both young women and men are presented in Table 2. For men only, possibility of deciding the work pace, possibility of taking short breaks, possibility of deciding when to perform a work task, and varied work showed unadjusted associations with excellent work ability. For men with high work demands, all of these variables except for possibility of taking short breaks also showed unadjusted associations with excellent work ability.

The variables included in the multiple models are shown in Table 3. For women, the exposure variable possibility of deciding working hours was included in the multiple model I with p = 0.08. In the multiple model II the same exposure variable was included; however, p fell to 0.1. Possibility of deciding the work pace was included in both these models only to adjust the models and thus should be viewed as a confounder. In the multiple model III for women with high work demands, both possibility of deciding working hours and possibility of deciding the work pace were included, however, with p = 0.1 and p = 0.2, respectively. In the multiple model IV for women only possibility of deciding working hours was included with p = 0.046.

Model I for the study sample of women included neither of the specified confounders (high work demands and educational level), but model II for the study sample of women included high work demands, Table 3. Further, model III for the subsample of women did not include the educational confounder, but model IV for the subsample of women did.

Other exposure variables were included in the multiple models for men. Having reported varied work was included in the two multiple models for all men, I and II, and also for men with high work demands, models III and IV, all with p < 0.05. Possibility of deciding when to perform a work task was included in both models I and III, with p = 0.097 and p = 0.099, respectively. In model II, this exposure variable was included with p = 0.1, and in model IV, p for this variable was 0.049. Possibility of deciding the work pace was included in models I, II, and III, with p = 0.1, p = 0.1, and p = 0.2, respectively. Finally, the variable possibility of taking short breaks was included, with p = 0.2 for all men in model I. In model II this exposure variable was only included to adjust the multiple model, as was possibility of deciding the work pace in model IV, and these should therefore be seen as confounders.

Similar to women, shown in Table 3, model I for the study sample of men included neither of the specified confounders (high work demands and educational level), and model II for the study sample of men included high work demands. Further, model III for the subsample of men did not include the educational confounder, but model IV for the subsample of men did, also in similarity with women.

The size of the effect of recovery opportunities at work on excellent work ability for young women is shown in Table 4 and for young men in Table 5. The largest effects were found for men who reported varied work (PR = 1.3 in all models), for men with high work demands with possibility of deciding when to perform a work task (PR = 1.3, model IV), and for women with high work demands who reported possibility of deciding working hours (PR = 1.2, models III and IV).

The most distinct finding in the current study was that young men with varied work seemed most likely to experience excellent work ability. The recovery obtained from varied work among the men can probably include both the possibility to take breaks, which has been shown to maintain performance among adult workers [9], and the possibility of deciding the work pace, both of which are in line with a theoretical framework of recovery in relation to variety in the job setting [14]. A review of studies in work physiology has clearly shown that variation in physical workload has a significant influence on recovery in relation to musculoskeletal disorders [25]. Creating varied work content could therefore contribute to promoting excellent work ability in young men, especially since younger workers tend to have more monotonous work than older workers [2].

The association between varied work and excellent work ability was not, however, any stronger for young men with high work demands. This indicates that varied work is important for excellent work ability, independent of the level of work demands. Further, the different results for varied work in women and men might be explained by gender segregation of different work tasks in the labour market [26], with different contributions to variation of work postures and work movements. Even in the same occupation it is known that work tasks often differ between women and men [27], and men are therefore more likely able to create variation in their work.

For men with high work demands, the possibility of deciding when to perform a work tasks was also found to be associated with excellent work ability. This recovery opportunity is included in the concept “global worktime control”, associated with job satisfaction, but not with performance or productivity [8]. Hence, the result of work ability in the current study has not, to our knowledge, been presented earlier.

Enabling women with high work demands to decide their own working hours could be a promising way to promote excellent work ability. This result is in line with earlier studies among adult workers. Having flextime has been shown to have an impact on job satisfaction [8] and a positive effect on productivity [10], although previously reported results were not specific for young women. Adult women with poor health and the opportunity to adjust their worktime were prospectively associated with increased work ability and return to work [28]; however, a large part of that study’s sample had poor work ability at baseline, which casts doubt on the appropriateness of a direct comparison with results of the current study. As no other recovery opportunities at work were found in the present study to be associated with excellent work ability among young women, other external factors such as duration and/or quality of sleep and/or relaxation between work-days may, in our opinion, be more important to their maintaining excellent work ability. Consequently, the situation could be more complex for young women, and possible important recovery opportunities were not taken into account in this study.

Obvious strengths of this study include its large population sample of a group not earlier investigated, but it also has limitations. The high proportion of young workers reporting excellent work ability and reporting high work ability in general might have muted the contrasts between exposed and unexposed groups and at least in part account for the small effects that were found. Furthermore, the limited scope of taking only internal recovery into account could make interpretations of the results uncertain, but probably contributes to a clearer focus.

The possible cohort effects in the current study may be an additional weakness as the sample was selected from different years. This sampling method of using surveys from three subsequent years was selected to obtain large groups for a study sample stratified for gender and further divided into subsamples with high work demands, while retaining enough power in the statistical analyses. Although the general work environment did actually change over the six years of the study [2], there is no reason to assume that possible associations between recovery opportunities at work and excellent work ability should also have changed substantially.

Excellent work ability can be measured in different ways. In the present study, as in earlier studies [3, 16], the cutoff for excellent work ability was 10 on the WAS. However, scores of 9–10 on the WAS have also been used for young workers [13, 29], in circumstances where choosing a score of 10 would have resulted in too few cases. Because self-rated work ability was high in the study sample, 10 on the WAS was considered appropriate for excellent work ability.

The questions in the telephone interview and the questionnaire were validated by Statistics Sweden, as occupational demands can be difficult to assess by self-reported exposure [30]. The validation procedure for these questions has been carefully described [15], and the questions used in the self-reported questionnaire were found to give reliable descriptions of actual work environments and conditions. Following that validation study, further work to increase the validity and reliability of the questions was also undertaken by Statistics Sweden; however, this work has not been carried out to a large extent since 2009, so the questions set for the three cohorts in the present study were nearly identical in formulation.

The choice of age group in the study sample warrants discussion. The upper age limit of 29 follows the limit for young workers used by Statistics Sweden, although their group starts at 16 years old. The lower limit of 18 years was chosen because most young adults aged 16–18 years continue in high school, and we wanted to examine young people at work.

The main limitation of the present study is the cross-sectional study design, which hinders interpretation of possible causal associations. Despite this limitation, however, the study has some obvious strengths. The design of a population-based register study is a broad attempt to capture the topic among a large group of young employees, and the well-described method selected for building the multiple models can be seen as an advantage. Furthermore, the research topic is, to our knowledge, poorly investigated in young workers, despite its possible importance to a sustainable working life for this young group.

The results of this study could be used when planning organizational actions to promote excellent work ability among young workers. Varied work might contribute to excellent work ability in men and plausibly a healthier workplace for all employees [31, 32]. Also, facilitating worktime control could be one way to promote excellent work ability, especially for women with high work demands, though the awareness of other opportunities for recovery at work and in leisure may be more important. Further studies, longitudinal or qualitative, are greatly needed to explore how recovery opportunities through the workplace can contribute to excellent work ability among young working women and men.