Study population

This study was the first nationwide survey of civil servants workplace health and used multistage stratified random cluster sampling based on proportional probabilistic sampling (PPS). The survey was launched by Taiwan’s Health Promotion administration (HPA) to assist with initiating an intervention program to improve worker’s health and the work environment. The results are to act as a baseline to evaluate the efficacy of the intervention program for civil servants. This study obtained ethical approval by the Institutional Review Board in the China Medical University Hospital (CMUH105-REC3-091). Participants were also informed that the data would be handled confidentially. Information about this study was sent to governmental institutions to invite workers to participate in our survey. Once agreed to participate, civil servants filled out informed consent and completed a self-report online questionnaire. Our study only enrolled participants aged between 40–60 years old. The overall response rate was 35.8%. Reasons for non-response included vacation, requested time off, or having insufficient time to fill out the questionnaire, and they do not affect the validity of the data. Our results validated the reliability of the web-questionnaire with a small group of 440 civil servants who were selected and completed both an in-person and web questionnaire. We obtained findings with high consistency from two types of interviews, which indicated that our results from the web-questionnaire were reliable. In addition, we obtained consist results from both high and low response groups, and there were no significant differences in the findings from the two response groups.

Measurement

The web-questionnaire included demographic data, anthropometric measurements (body height and weight), lifestyle habits (tobacco smoking, alcohol consumption, betel nut chewing, duration of sleep, and frequency of physical activity), and history of diseases. Our study population was categorized into smokers and never smokers. Non-smokers or former smokers were originally included in the category of smokers because the sample size of former smokers was quite small. In this study, ETS exposure was defined as living with family members who smoked (ETS at home) or working in the same office as colleagues or managers who smoked (ETS in workplace). Effort-reward imbalance (ERI) was used to assess the level of perceived psychosocial work stress [10]. ERI was calculated using three items that measured ‘effort’ and seven items that measured ‘reward’ [11]. The ERI ratio was computed using the sum of the scores for efforts as the numerator (Efforts) and the sum of the scores for rewards as the denominator (Rewards) with an ERI ratio > 1 representing high perceived psychosocial work stress in the workplace. Cronbach's alpha coefficients for each scale in the ERI model ranged from 0.65–0.88, which was consistent the Siegrist’s study (i.e. 0.61–0.91) [11].

The web-questionnaire had three questions regarding hyperlipidemia: “Have you suffered from hyperlipidemia?” (self-perceived), “Has the disease been diagnosed by physicians?” (physician diagnosis) and “Is the disease being treated and followed up by a physician?” (ongoing treatment). We used the physician’s diagnosis as the measurement for hyperlipidemia. Overweight and obesity were defined using body mass index (BMI) measured by weight and height ratio (kg/m²). Based on Taiwan’s HPA guideline (2012) for overweight and obesity for Taiwanese adult men and women, BMI was categorized as follows: 18.5–24 (healthy), 24.1–27.0 (overweight), 27–30 (mild obesity), 30–35 (moderate obesity), and 36 and higher (severe obesity).

Statistical analysis

SPSS 24 was used to analyze the data. Univariate analysis was used to examine the difference between the two ERI groups on demographic information. Hyperlipidemia was associated with the following risk factors: ERI, lifestyle, overweight, obesity and levels of BMI based on univariate analysis. Multiple logistic regression adjusted for gender, age, education level, income, and marital status was used to assess ERI levels associated with hyperlipidemia for both groups of ETS exposure, obesity, and lifestyle habits. Interaction effect of ETS exposure and ERI on risks of hyperlipidemia in the non-obese and obese groups were assessed using multiple logistic regression adjusted for gender, age, education level, income, marital status, and smoking. Sensitivity analysis was used to examine the interaction effect using different measurements for hyperlipidemia. The interaction effect of ETS exposure and ERI on hyperlipidemia was measured by relative excess risk due to interaction (RERI), attributable proportion due to interaction (AP) and synergy index (S). A resulting confidence interval of RERI > 0 or AP > 0 or S > 1 indicated an additive interaction (available at website with http://www.epinet.se.).

Risk factors for hyperlipidemia

Available data have shown that the prevalence of hyperlipidemia in Taiwan has rapidly increased due to changes in lifestyle habits and dietary behaviors, which are significant contributing factors of CVDs. Based on the second Nutrition and Health Survey for Taiwanese men from 2005 to 2008, the prevalence of hypercholesterolemia, defined as total cholesterol (TC) ≥240 mg/dL, was 13%, and the prevalence of hypertriglyceridemia, defined as triglyceride (TG) ≥ 200 mg/dL, was 21% [12]. In this study, 4.4% of civil servants who responded with self-reported hyperlipidemia were diagnosed subsequently by physicians based on high cholesterol or high triglyceride levels from annual routine medical examinations. Our results are consistent with previous studies showing that risk of hyperlipidemia were significantly associated with tobacco smoking (OR = 1.84), alcohol consumption (OR = 1.67), and betel nut chewing (OR = 2.84) [13–15]. Evidence indicated that smoking might disrupt lipid and lipoprotein metabolism, elevating plasma triglycerides, cholesterol, and low-density lipoprotein (LDL) levels and decreasing high-density lipoprotein (HDL) cholesterol level [13, 15]. Taiwan has a high rate of betel nut chewing, which stimulates one's appetite by inhibiting GABA receptors and increases the risk of obesity and occurrence of CVDs [16]. A study in Taiwan showed that the risk of obstructive coronary artery disease (CAD) was significantly associated with areca nut consumption (OR = 3.5) and showed additive interactions between areca nut use, cigarette smoking, and dyslipidemia after adjusting for other covariates [17].

Work-related stress in the workplace, including precarious work, job insecurity, overcrowding, high workload demand, inflexible operational systems, physical inactivity from extended work hours, insufficient leisure-time, unbalanced diet, depression, and anxiety, ultimately leading to CVDs [18]. Exposure to long-term stressful work can lead to high cholesterol levels, triglycerides, hematocrit, fibrinogen, and blood fluidity, and increased coagulation and platelet activity, all of which can act as “triggers” of CVDs [19, 20]. However, previous studies reported an inverse association between high levels of chronic stress and cardiovascular reactions. These studies did not consider various contributing factors of CVDs such as unhealthy lifestyle or mediating factors such as social support and obesity [21, 22]. A study for young male firefighters indicated that social support may help prevent chronic stress by enhancing recovery [23]. As hyperlipidemia is often a precursor to CVDs, the focus on it was more suitable for the bounds of this study because employees suffering from CVDs may have retired or left the workforce, and therefore, were no longer available for further analysis. Our results showed higher ERI and thus higher risk of hyperlipidemia for alcohol (OR = 1.82) and betel nut (OR = 1.92) consumers compared to non-consumption of alcohol and non-betel nuts (OR = 1.53 and OR = 1.54, respectively). However, there was an inverse relationship for smokers (OR = 1.49) and non-smokers (OR = 1.58). It is speculated that the association between the risk of hyperlipidemia and ERI may be attenuated by the effects of smoking. Notably, for alcohol drinkers and betel nut chewers, there was higher perceived work stress, which can cause a higher risk of hyperlipidemia.

ETS, perceived work stress and hyperlipidemia and CVDs

Emerging evidence indicates a linkage among smoking and both atherosclerosis and cardiovascular disease. We hypothesize that combined exposure to high ERI and ETS can cause hyperlipidemia and increase the risk of CVDs through oxidative stress and mitochondrial damage [24]. For civil servants, ETS exposure and perceived work stress often occur together, yet their combined effects on hyperlipidemia and CVDs development currently remain unclear. Even though there is a smoke-free policy in Taiwan, civil servants may still be particularly vulnerable to ETS or THS exposure in the workplace due to their responsibility to serving citizens who smoke or are exposed to ETS. Moreover, other challenges of external competition and governmental workforce downsizing in Taiwan have led to higher perceived stress, poor sleep quality and reduced sleep hours, unbalanced diet, and physical inactivity. Consequently, civil servants have a higher risk of weight gain and hyperlipidemia due to stressful work. Also, our results exhibited interaction effects of ETS exposure and high ERI on the risk of hyperlipidemia adjusted for covariates, regardless of obesity and smoking. More critically, there were higher interaction effects that occurred in the non-obese (OR = 1.84) and non-smoking (OR = 1.89) groups, compared to the obese (OR = 1.62) and smoking (OR = 1.57) groups. The interaction effects of ETS exposure and high ERI on hyperlipidemia were dominated in the non-obese and non-smoking groups. Therefore, the adverse effects of ETS and high ERI level may explain, in part, the increased risk of hyperlipidemia, which may then cause the progression of CVDs or other metabolic diseases.

Previous studies [13, 25, 26] indicated that cigarette smoking is a contributing factor on HDL levels and functioning, revealing that each 1mg/dl decrease in plasma HDL cholesterol concentration is associated with a 2% increased risk of CVDs for men and a 3% increase for women. In a study [27] of developing countries, there was a strong positive association between ETS exposure and ischemic HD (main relative OR = 1.17–2.36) and between ETS and stroke [HR (hazard ratio) = 1.41–1.49]. Approximately 30% of heart disease (HD) risk is attributed to exposure to ETS through possible biological mechanisms related to platelet activation, endothelial dysfunction, inflammation, atherosclerosis development and progression, increased oxidative stress, decreased energy metabolism, and increased insulin resistance [28]. The mechanism of developing CVDs due to ETS exposure is linked to an increase of the proatherogenic effect of dyslipidemia. However, inconsistent findings from a study [29] reported that childhood ETS exposure was significantly associated with stroke (OR = 1.66, 95% CI = 1.29–2.13) but not CVD (OR = 1.15, 95% CI = 0.82–1.59). Limitations of the study included not adjusting for the other risk factors, retrospective self-reporting of childhood ETS exposure, and causality of exposure to diseases occurring at least 27 or more years apart [30]. Subjects with high-perceived work stress may use various negative coping strategies, such as taking drugs or using alcohol to feel better and to avoid unpleasant thoughts and painful emotions. Similarly, high levels of job strain and job demands were associated with a higher likelihood of smoking in an observational study in Finland [31]. Most of the previous studies neglected to elaborate on the combined effects of ETS exposure and perceived stress in the workplace on the risk of CVDs, metabolic diseases, and depression. Our results have indicated that the risks of hyperlipidemia were significantly associated with lifestyle habits (smoking, alcohol consumption, betel nut chewing), ERI, and weight gain for middle-aged civil servants. Moreover, both exposure to ETS and perceived stress may contribute to a synergism effect towards the risk of hyperlipidemia, especially for the non-obese and non-smoking groups.

Since the government implemented the smoke-free workplace policy in 2009 (which included federal workplaces), smokers in civil servants may have quit smoking or immediately reduced their cigarette consumption. Civil servants may have decreased ETS exposure in worksite between coworkers, but they may still have ETS exposure from citizens carrying smoke absorbed in their clothing or on their skin. Therefore, there needs to be a strict implementation of a smoke-free policy in the workplace as well as enforcement of smoking cessation and frequent remediation of THS on office walls. In addition, stressful work for civil servants should be reduced by modifying rigid bureaucratic systems, providing social support from colleagues and management, and encouraging healthy lifestyle habits.

Strengths and limitations

Strengths of this study include using the first nationwide study of civil servants in Taiwan to examine the interaction effect of ETS exposure and perceived work stress on hyperlipidemia. The representative population was selected by PPS sampling and consistent results were validated using web- and paper-based questionnaires. Our results can be used to initiate an intervention program for high risk groups and consequently used for follow-up evaluations on the efficacy and effectiveness of the program. However, there are several limitations in the study. First, it was hard to establish causality based on the cross-sectional design. Second, complete risk factors for hyperlipidemia, such as daily diet behaviors and family history, were not collected for each participant. Although our results measured various lifestyle behaviors including cigarette smoking, betel nut chewing, sedentary time, physical activity, and sleep hours, we were unable to verify the accuracy of these self-reported behaviors. Third, the onset of hyperlipidemia was also self-reported and there could be possible recall loss from participants. Based on the sensitivity analysis for the three hyperlipidemia measurement types, our results indicated no difference on the interaction effects of ETS exposure and high ERI. In addition, our results may have a “healthy worker effect”, which can be attributed to early retirement or layoffs, and yield an underestimation in our findings. Fourth, the source and differences of workplace stress between public and private sector employees make comparing the exposures to ERI and ETS difficult and their relationship to hyperlipidemia. The limitations of this study’s population reduce the variability of these factors to produce a stronger case for the association. While the source and cause of ERI and ETS will differ between the subjects of this study (civil servants) and the general population, the findings regarding the relationship between both ETS and ERI exposure on hyperlipidemia are still applicable. Finally, the mechanisms responsible for the interaction effect of ETS exposure and perceived work stress on hyperlipidemia have not been clearly elucidated.