We conducted a cross-sectional analysis of the baseline data of a workplace cohort study in Singapore. Details of the cohort study design have been reported elsewhere [19]. Briefly, the study comprised 464 full-time employees from 4 companies. Employees from the transport industry, cooling plants, and universities were recruited via workplace posters and emails. They were eligible for participation if they fulfilled the following criteria: at least 21 years old, English-speaking, working for at least 4 hours per day, and not pregnant. Recruitment and baseline assessment were conducted from August 2017 to March 2018.

We summarized continuous variables as mean±standard deviation, and categorical variables as frequency and percentage. Continuous data were checked for skewness, and skewed variables were presented as median and interquartile range. To identify factors associated with PCS and MCS, a generalized linear model was fitted with a normal distribution and an identity link function using a robust standard error in the following hierarchical fashion:

Model 1: included socio-demographic characteristics; Model 2: retained variables of model 1 with a p-value < 0.20 and included health factors/lifestyle behaviours; Model 3: retained variables of model 2 with a p-value < 0.20 and included occupational factors; Model 4 (final model): retained variables of model 3 with a p-value < 0.20.

A two-tailed p-value < 0.05 was considered to indicate statistical significance. All statistical analyses were conducted using Stata version 15.0 (StataCorp., College Station, TX, USA).

The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board (IRB) of Nanyang Technological University (NTU) (IRB-2015-11-028). The study participants provided written informed consent prior to the commencement of data collection.

We found that age and ethnicity were significantly associated with MCS and PCS, respectively. Older workers reported higher MCS scores, which is consistent with the findings of other studies that older age is associated with better mental health outcomes [26]. Workers in the older age group benefit from better mental health when highly engaged in their work [27]. In an earlier longitudinal study conducted on middle-aged and older Singaporeans, work engagement was shown to promote improved mental well-being, as employment provides opportunities for social interactions and cognitive engagement [28]. In contrast to the younger age group, older workers may have built up resilience in managing changes or stressors in the work environment from their years of experience in the workplace, enabling them to enjoy better mental well-being [27].

We also observed that Malay workers were more likely to report lower PCS scores than their Chinese counterparts. In a national cohort study conducted in Singapore, participants of Chinese ethnicity reported significantly higher PCS than those of other ethnic groups (i.e., Malay and Indian). However, in that study, the effect of ethnicity on physical HRQoL in PCS was no longer significant after adjusting for other variables. Previous studies in Singapore have reported ethnic differences in some health conditions and behaviours. For instance, individuals of non-Chinese ethnicity were more likely to experience migraines [29] and type 2 diabetes [30], meaning that they are more likely to report poorer physical well-being. There may be cultural differences in help-seeking behaviours amongst the different ethnicities, especially in the way they perceive physical symptoms and seek professional help for diagnosis or treatment [31]. For instance, Malays may perceive pain conditions as involving spiritual resilience, whilst Chinese may seek alternative treatments like acupuncture or traditional Chinese medicine to treat their symptoms. A broad range of factors mediate relationships between ethnicity and health, such as cultural beliefs and behaviour, social networks, family stress, and personality characteristics [32]. Thus, it is crucial to note that the use of PCS as a measure of physical HRQoL may not be sensitive to the complex nature of ethnic or cultural differences in the current population.

Of all the lifestyle and health factors measured, only overweight or obesity significantly contributed to physical HRQoL in our working population. This is aligned with previous research in Singapore that has identified associations between a higher BMI and lower physical HRQoL [11]. Further studies have also supported associations between obesity and the development of musculoskeletal disorders, which result in back, joint, or overall muscle pain that may impede employees’ ability to perform work-related physical tasks [33]. Whilst it is expected that participants who were overweight/obese would report lower scores on physical HRQoL, it is also worth exploring other aspects of their behaviour that may contribute to their lower perceptions of physical well-being. Notably, treatment-seeking overweight or obese individuals are more likely to report lower HRQoL, in the domains of bodily pain, general health, and vitality, than their non-treatment seeking counterparts [34]. Their lower perceptions of well-being could be a result of greater body dissatisfaction and more effort should be made to promote healthy lifestyles to support a positive body image for these individuals.

In our current sample, workers who were overweight or obese reported significantly lower scores on subscales measuring physical functioning and physical role limitation (data not shown). Similarly, a workplace study found that employees in the higher BMI categories (≥ 35 kg/m²) were observed to experience significantly greater difficulty in performing work-related physical tasks [35]. This was estimated to result in a 1.2% health-related loss in productivity as compared to employees in the lower BMI categories. In general, productivity loss as a result of work limitations in overweight or obese workers could be attributed to increased absenteeism, which refers to the inability to work, or increased presenteeism (i.e., the inability to work at full capacity, while being present on the job) [36].

Consistent with extant research, worse sleep quality was significantly associated with poorer physical and mental HRQoL. Studies of people with insomnia using the 36-Item Short-Form Survey have consistently reported impairments in diverse aspects of HRQoL, when compared to those with better sleep quality [37]. The association between these 2 variables is consistent despite differences in the classifications and severity of insomnia or poor sleep. There are numerous health consequences of the lack of good sleep, involving both the physical and emotional aspects of functioning [38]. Shorter sleep duration with sleep disturbances was previously identified to have an effect on coronary heart disease, and shorter sleep duration was associated with an increased risk of cardiovascular complications, including stroke [39]. Sleep disturbance and poor sleep quality can, amongst other effects, result in decreased motivation and attention deployment and overall lower psychosocial well-being [40]. This is also relevant to performance at work, as poor sleep quality is associated with concentration and organization. A stress-health model describing the inter-relationships of sleep, stress, and illness incorporates both poor sleep and stress as contributors to allostatic load, which refers to the repeated exposure and adaptation to stress that results in physiological strain [41]. This would result in illness, which in turn contributes to stress and poor sleep, thereby perpetuating a cycle of poor sleep and declining mental and physical health.

Workers who experienced varying degrees of stress at work in our study reported significantly poorer mental HRQoL. According to a recent well-being survey, 92% of working Singaporeans reported being stressed at work, which is well above the global average of 84% [42]. With Singapore’s rapid industrialisation and economic growth, the pressure to meet higher expectations for productivity and efficiency is inevitable. Other related studies on working professionals found strong negative associations between occupational stress and HRQoL [5]. Workers with lower work autonomy and control might perceive higher levels of stress, which is often observed amongst less professionalized jobs [13]. A person’s natural response to stress is multidimensional. Physiologically, activation of the autonomic nervous system, consisting of the hypothalamic-pituitary-adrenal and sympatho-adrenomedulllary axes, makes the body vulnerable to diseases [43]. At the cognitive level, to cope with the combination of high demands and low control, an individual may maintain a state of high sustained effort, known as strain [44]. Increased job strain has been shown to have substantial negative influences on workers’ mental health, in addition to the loss of productivity from not being able to work at optimum conditions [6].

In our study, the number of years spent at the company was significantly associated with physical HRQoL, but not mental HRQoL. When adjusted for other variables, the association remained significant (p< 0.05), with more years at the company-related to better physical HRQoL. We hypothesized that employees working for a longer period at their current company are better at adjusting and adapting to changes in the workplace, and are therefore more resilient and may perceive better physical well-being. In a previous study of morbidly obese patients, being employed was a protective factor for physical and mental HRQoL. This was most marked in the subscales of physical functioning, physical role limitation, bodily pain, and emotional role limitation [45]. The social networks formed in the workplace, as well as being physically and mentally occupied at work, contribute to the positive observed associations of HRQoL.

The outcome of HRQoL may differ between occupational groups, as mental and physical characteristics vary, and non work-related factors (e.g., level of education or lifestyle) differ between groups [12]. In our study, occupation type was significantly associated with physical HRQoL, with desk job workers reporting higher PCS scores than those with non-desk jobs. However, this was no longer significant after adjustment for other variables, which could be due to the associations between occupation type and lifestyle or health-related behaviours such as physical activity or smoking status.

In light of the extent to which overweight/obesity is significantly associated with workers’ physical HRQoL, a health promotion programme initiated at the workplace, where individuals spend much of their time, may benefit the most for these workers. Health promotion programmes at the workplace ensure that a significant proportion of the adult population is reached within a relatively controlled setting. Weight loss programmes at the workplace have been shown to improve physical HRQoL, particularly in the domains of physical functioning, physical role limitation, and general health perceptions [46]. For such programmes to be effective, they need to be accessible and easily adhered to, meaning that they must be low-cost and provide strong socio-emotional support for participants. Workplace policies and strategies should also be aimed at promoting work-life balance and managing the emotional and physical responses to specific job demands so as to ensure the holistic (i.e., physical, mental, and social) well-being of employees. A holistic approach enables programmes to address workers’ experience of stress at work with the aim of improving their overall mental HRQoL. For instance, workplaces could focus on building a supportive management system, reducing unfavourable work conditions and promoting growth, autonomy, and opportunities for employees [47]. This would help to build resilience in workers and equip them with strategies to cope with stressors in the work environment, thus improving their perceptions of their functional ability and general well-being. For employers, the possibility of increased productivity could be a strong incentive to embark on such initiatives.

Some strengths of the current study include the use of a large and representative sample of workers from different occupations and work environments. We also included an extensive list of health, lifestyle, and work-related variables that have been related to poor health outcomes in working populations, and there were little missing data for the variables measured.

There are also limitations present in this study. The cross-sectional study design does not allow for the examination of causality between the independent variables and HRQoL. Moreover, reverse causality cannot be ruled out. Longitudinal data would be necessary to understand the direction of causality. There is also a possibility of self-selection bias in the recruitment of participants. Employees who are experiencing poorer HRQoL may have been more motivated to participate in the study than those who have better HRQoL. Therefore, the observed associations may not be generalisable to the wider working population in Singapore. Although the results are based on different occupation types, the group of non-desk job workers was too small for a stratified analysis to identify differences in risk factors for different occupation types. In addition, physical and mental HRQoL were assessed by a self-reported questionnaire. Assessing physical and mental HRQoL with validated and culturally adopted tools may be necessary [48]. This could provide better insights into the well-being or quality of life of the working population. Furthermore, there might be additional variables that were not adjusted in the analysis. We assessed stress at work, but other work-related factors (e.g., decision latitude, job demands, or social support) have been shown to be related to HRQoL [49]. For workspace location, we tested aboveground versus underground locations. However, there are other factors in the working environment, such as workspace design and functional comfort, that could also affect workers’ quality of life [50].