Association between health literacy and behaviors among shift workers: an observational cross-sectional study with mediation analysis

Abstract

Introduction

Owing to globalization and the 24-hour service culture, shift work has become a common working style, with 17% and 16% of workers engaged in shift and night work, respectively, in Japan.¹ Previous studies have reported that shift work is a risk factor for cardiovascular disease (CVD), with meta-analyses showing that shift work increases the risk of CVD development.^2-4 Unfavorable health behaviors, including smoking, poor sleep duration and quality, adiposity, and metabolic status, were identified as mediators between shift work and CVD.⁵ Regarding adiposity and metabolic status, lower physical activity levels and low vegetable consumption were noted as mediators.⁶ The findings on dietary habits are generally similar, with shift workers showing irregular and more frequent meals, increased nighttime snacking or eating, and lower vegetable and fruit consumption than daytime workers.^7-9 Moreover, several articles have reported evidence linking shift work with less physical activity and increased rates of smoking.^10,11

Clarifying factors related to the differences in health behaviors between shift and daytime workers would be useful in developing strategies for bridging the gap between the two. Health literacy (HL) is related to health behaviors. HL is defined as “cognitive and social skills that determine an individual’s willingness and ability to access, understand, and use information to promote and maintain good health”¹² and has an observed association with health behaviors.^13-15 Misalignment with family and social rhythms of shift workers¹³ may affect the access, understanding, and use of health information, which are components of HL, and may lead to differences in HL levels. Conversely, their irregular lifestyles may make it difficult for them to adopt health behaviors regardless of HL level. Some follow-up studies have shown increased smoking rates, decreased leisure time physical activity, and irregular eating habits of shift workers following new engagement in shift work.¹⁴ However, to date, no studies have compared the HL levels of shift and fixed daytime workers. Additionally, no studies have investigated whether HL mediates differences in health behaviors between shift and daytime workers.

Therefore, this study aimed to investigate whether a difference in HL exists between shift and fixed-day workers and clarify whether a mediating role of HL in the relationship between health behaviors and shift work exists in a large-scale manufacturing company.

Methods

Ethical consideration

The Kanazawa Medical University Medical Research Ethics Review Board approved this cross-sectional study (approval number: I431). Moreover, this study was approved by a committee composed of representatives of the target company.

Target population and survey methods

This study employed a cross-sectional design, with the target population all being employees (n = 8215) in a light metal product manufacturing company in Toyama Prefecture, Japan. The target company had an average ranking for occupational safety and health. In January 2020, using a self-administered questionnaire, the Japanese version of the European Health Literacy Survey (HLS-EU)-47, we surveyed job characteristics, health behaviors, and HL¹⁶. This study was part of a longitudinal survey, and all participants provided written informed consent after receiving a document describing the entire study program. Information that could identify an individual during data processing was converted into a pseudonym by a person in charge of the business, with the researchers receiving de-identified data for analysis.

Study participants were manual workers in the target company, whose jobs involved either casting, hot rolling, cutting, processing, or working on the assembly line. They had relatively the same job position, and none of them were managers. The work schedule (shift work or fixed-day work) and working hours per week (<30, 30-34, 35-39, 40-48, 49-59, or ≥60 hours) were self-reported. The shift schedule in the target factory included a rotating 3-shift schedule and a 2-shift schedule. Two-thirds of the 3-shift workers had a noncontinuous shift system (5-day, night, and evening shifts), whereas one-third had a continuous shift system (3- or 4-day, night, and evening shifts).

Regarding personal attributes, we asked about educational history (junior high school, high school, vocational school, junior college/technical college, university, or graduate school graduates) and marital status (married, single, or divorced/widowed).

Regarding health behaviors, we asked about the frequency of weekly exercise (no weekly exercise or sports, light exercise at least once a week, intense exercise once or twice a week for ≥20 minutes, intense exercise three times or more a week for ≥20 minutes), smoking history (currently smoking, quit smoking, or no smoking history), frequency of general food intake per day or week, and the frequency of eating breakfast (almost every day, 4-5 d/wk, 2-3 d/wk, or rarely). Furthermore, we asked separate questions about eating vegetables, either green and yellow vegetables or other vegetables, and the daily frequency of tooth brushing (less than twice a day or more). To evaluate HL, the Japanese version of the HLS-EU-47¹⁶ was used. This scale was developed by the HLS-EU and encompassed 47 items related to 3 domains (health care, disease prevention, and health promotion) and 4 competencies (obtaining, understanding, evaluating, and utilizing health information). A possible score ranged from 0 to 50 points; the higher the score, the higher the HL.

We received survey responses from 6962 individuals (response rate, 84.7%). Of these respondents, 5915 (72.0%; 4058 males and 1857 females) had no errors in the analysis parameters. In the target company, females rarely engaged in 3-shift schedules. Among 4058 male respondents, 1906 were nonmanual workers; therefore, the final analysis included 2152 male manual workers aged 18-64 years (961 and 1191 fixed-day and shift workers, respectively) (Figure 1).

Figure 1

Flow diagram of the study.

Open in new tabDownload slide

Analytical methods

The following conditions were defined as favorable habits: “performing leisure time exercise more than once a week (performing exercise),” “currently no smoking (nonsmoking),” “having breakfast almost every day (having breakfast),” “brushing teeth at least twice a day (brushing teeth),” and “eating green and yellow vegetables at least once a day (eating vegetables).”

HL scores were categorized into the following 4 levels according to the European Health Literacy Project 2009-2012¹⁷: 0-25, 26-33, 34-42, and 43-50 points for inadequate, problematic, sufficient, and excellent, respectively. Educational attainment (lower than university education or university education or higher), marital status (married or single), and working hours (48 h/wk< and ≤48 h/wk) were other factors considered.

The χ² test for categorical variables and analysis of variance were employed to compare sociodemographic characteristics between fixed-day and shift workers. Analysis of variance and analysis of covariance, with age as a covariate, were used to compare HL between the 2 groups. After stratification by age, the χ² test was used to compare HL distributions between fixed-day and shift workers.

Before multivariate logistic regression analysis, the percentages of favorable habits (performing exercise, nonsmoking, having breakfast, brushing teeth, and eating vegetables) were compared between the 2 groups according to each of the following factors: work schedule, working hours, educational attainment, marital status, and HL group.

After adjusting for age, HL, educational attainment, marital status, and working hours, the odds ratios (ORs) of favorable habits in shift workers compared with those in daytime workers were subsequently calculated using multivariate logistic regression analysis. Furthermore, after adjusting for age, work schedule, educational attainment, marital status, and working hours, the ORs of favorable habits in the HL groups were calculated.

Logistic regression structural equation models were used to analyze the mediating role of HL in the relationship between shift work and health behaviors.¹⁸ A schematic presentation of the framework used for examining the relationship between shift work and health behaviors is shown in Figure 2. The upper part of the framework shows the total effect (c) of shift work on health behaviors adjusted for age, marital status, educational attainment, and working hours. The lower part depicts the mediation model of the indirect effect of HL (a, b) and the direct effect of shift work (c′) on health behaviors. The 95% CIs were calculated on the basis of 5000 bootstrap resamples.

Figure 2

Multiple mediation model.

Open in new tabDownload slide

Statistical analysis was performed using Statistical Package for the Social Sciences (version 27, IBM), and PROCESS version 4.3 of SPSS was employed for the mediation analysis. The level of statistical significance was set at P = .05.

Results

The demographic characteristics and HL of the participants, grouped according to work schedule, are summarized in Table 1. The mean age of the participants was 38.1 (SD = 13.2) years, with shift workers being significantly younger than daytime workers (P < .001). Daytime workers had a significantly higher proportion of participants with a university education or higher or being married than shift workers (both P < .001). The 2 groups showed no differences in the proportion of participants who worked >48 h/wk (P = .073). The mean HL of all participants was 26.3 (SD = 7.9). Analysis of variance showed a significant difference in HL between fixed-day (mean = 25.8; SD = 7.6) and shift workers (mean = 26.8; SD = 8.1); however, the age-adjusted mean was not significantly different (P = .348). Following stratification for age group, HL distributions were not significantly different between fixed-day and shift workers (P = 0.181 for 18-39 years; P = .821 for 40-65 years). Among the participants aged 18-39 years, the largest proportion of participants (41.6%) had problematic HL followed by inadequate, sufficient, and excellent HL. Among the participants aged 40-64 years, the largest proportion (53.5%) belonged to the inadequate HL, followed by problematic, sufficient, and excellent HL.

The prevalence of favorable health behaviors grouped according to work schedule, working hours, HL, and demographic characteristics is presented in Table 2. Shift workers had a significantly lower prevalence of nonsmoking, having breakfast, or eating vegetables than daytime workers, with statistically significant differences (all three P < .001). A significantly higher prevalence of eating vegetables was observed in the group working >48 h/wk than in the group working ≤48 h/wk (P = .008). A significantly higher prevalence of performing exercise, nonsmoking, or brushing teeth was noted in the group with a university or higher education than in the group with lower educational attainment (P = .009, .021, and .016, respectively). The group with married participants had a significantly higher prevalence of all 5 behaviors than the group with unmarried participants. HL was associated with the prevalence of performing exercise, brushing teeth, and eating breakfast (P < .001, .012, and <.001, respectively).

The relationship between work schedule and health behaviors determined by multivariable logistic regression analysis is presented in Table 3. After adjusting for age, HL, educational attainment, marital status, and working hours, the ORs and 95% CIs of shift workers, with daytime workers as the reference group, were 0.85 (95% CI, 0.70-1.02), 0.68 (95% CI, 0.56-0.80), 0.63 (95% CI, 0.52-0.76), 0.79 (95% CI, 0.65-0.95), and 0.58 (95% CI, 0.48-0.70) for performing exercise, nonsmoking, having breakfast, brushing teeth, and eating vegetables, respectively. A higher HL was significantly associated with performing exercise, having breakfast, and eating vegetables (P value for trend <.001, .01, and <.001, respectively).

Mediation analysis results, including the total (c), direct (c′), and indirect effects (a*b) of shift work on favorable habits calculated using multiple logistic regression analysis, are summarized in Table 4. The OR between shift work and HL (path a) was 1.04 (95% CI, 0.97-1.12). The ORs and 95% CIs between HL and health behaviors (path b) were 0.75 (95% CI, 0.67-0.84), 1.01 (95% CI, 0.90-1.13), 0.86 (95% CI, 0.76-0.96), 0.91 (95% CI, 0.81-1.02), and 0.80 (95% CI, 0.72-0.89) for performing exercise, nonsmoking, having breakfast, brushing teeth, and eating vegetables, respectively. The ORs and 95% CIs of the direct effects of shift work on favorable habits were almost the same as the total effects as follows: 0.85 (95% CI, 0.70-1.02), 0.67 (95% CI, 0.56-0.81), 0.63 (95% CI, 0.52-0.76), 0.79 (95% CI, 0.64-0.95), and 0.58 (95% CI, 0.48-0.70) for performing exercise, nonsmoking, having breakfast, brushing teeth, and eating vegetables, respectively. The indirect effects (ORs) of HL on performing exercise, nonsmoking, having breakfast, brushing teeth, and eating vegetables were 1.01 (95% CI, 0.99-1.03), 1.00 (95% CI, 0.99-1.01), 1.01 (95% CI, 1.00-1.02), 1.00 (95% CI, 1.00-1.02), and 1.01 (95% CI, 0.99-1.03), respectively.

Discussion

This study investigated whether a difference in HL exists between shift and daytime workers, and clarified whether a mediating relationship in the disparities of health behaviors exists between shift and daytime workers. No differences in the age-adjusted distribution of HL were observed between fixed-day and shift workers. Additionally, the difference in health behaviors between fixed-day and shift workers was not mediated by HL.

A lower prevalence of favorable habits was noted among shift workers than among fixed-day workers, except for performing exercise. Consistent with our results, previous studies have reported that shift workers are more likely have an unbalanced diet,^{6,  15-18} a high prevalence of smokers,^3,5, and that they are less likely to change their behavior.¹⁹ Furthermore, the results of performing exercise were inconsistent. The Finnish Public Sector Study showed that male white-collar workers with nighttime work schedules were more likely to be physically active but not female or blue-collar workers.²⁰ Other studies did not find any significant difference.^14,21 Whether or not individuals perform exercise during their leisure time is believed to be determined by several factors, including health awareness, the length of their leisure time, the amount of time spent on housework, and the degree of physical fatigue. These differences in individual factors may have contributed to the differences in the results.

The HL of the target population was similar to that reported by Nakayama et al²² and Kimura et al²³ for individuals in Japan but was lower than that reported for individuals in European countries.²⁴ HL was related to health behaviors except for smoking, independent of work schedule. The possibility that the irregular work system of shift work causes disharmony with the rhythm of society and hinders the acquisition and use of health information, that is, HL development, was denied. However, the irregularity of the life rhythm caused by the work pattern itself may make adopting healthy behaviors difficult. Follow-up studies have shown increased smoking rates after new engagement in shift work.¹⁴ Bekkers et al ²⁵ observed that individuals who moved from shift to daytime work changed their breakfast behaviors, and those who moved from daytime to shift work more frequently started smoking. These findings suggest that an educational program that increases HL would not narrow disparities between shift and daytime workers. This finding further suggests that workers’ working styles and environments must be considered when attempting to improve their health behaviors.

Several studies have reported that HL was not associated with smoking and having breakfast daily. A nationwide cross-sectional survey with a random stratified sampling design in the Netherlands by Svendsen et al²⁶ indicated that inadequate HL was strongly associated with inactivity and overweight but, to a lesser extent, with health behavior factors, including smoking and high alcohol consumption. The 2013 Health Information National Trends Survey also reported the lack of association between smoking and HL.²⁷ This finding suggests that environmental changes are more effective than improving individual HL in changing addictive habits, including smoking and drinking alcohol. However, the relationship between HL and having breakfast has not been clarified in previous studies.

This study revealed no difference in HL levels depending on work style and suggests that shift work is related to lifestyle habit formation. However, it had some limitations. First, as this was a cross-sectional study, distinguishing whether health behaviors existed before starting work or developed later was impossible. It has been reported that shift workers in social and health work have higher smoking rates even before they start shift work.²⁸ Second, as the company being investigated was relatively active in promoting health in the workplace, using a population and a high-risk approach, these activities may have affected our findings. Third, in this study, the definition of breakfast was unclear, which may have led to bias in this item. Breakfast is usually defined as the first meal eaten in the morning after sleep; however, night shift workers may be confused about which meal to consider as breakfast. Lastly, the evaluations of HL and behaviors were self-evaluations; consequently, information bias may be present.

Although limitations exist, this study suggests that the disparity in health behaviors between shift workers and fixed-day workers may not be attributed to differences in HL. Further research is needed to clarify the causes of these differences.

Acknowledgments

We thank Yuchi Naruse, Chiaki Okamoto, and Yuki Nakashima (Health Care Center, YKK Corporation) for scientific advice and for performing the questionnaire survey.

Author contributions

Y.M. conceived the ideas; Y.M., K.T., M.S., M.I., T.K., and H.N. designed the study. Y.M., M.S., and M.I. collected the data. Y.M., K.T., T.K., and M.S. analyzed the data. Y.M. and M.S. wrote the paper. Y.M., K.T., M.S., M.I., T.K., and H.N. reviewed the manuscript and accepted the final version to be published.

Funding

This research was supported by the Japan Society for the Promotion of Science, Grant Number 18 K10093.

Conflicts of interest

The authors have declared that no competing interests exist.

Data availability

The datasets generated and/or analyzed during the current study are not publicly available due to information management rules of the target company but are available from the corresponding author upon reasonable request.

References