Study population
This study was a secondary data analysis that utilized data obtained from the Fifth KWCS of 2017, conducted by the Occupational Safety and Health Research Institute (Approval No. 380002) [17]. The original data were collected from workers aged 15 years and above living in Korea through a 1:1 interview. The survey was conducted at the residences of the sample population from July to November 2017. Samples of the original data were extracted and stratified according to the characteristics of the administrative and enumeration districts using the 2010 Population and Housing Census as the sampling frame [17]. After selecting the enumeration district using probability systematic sampling, subsequent systematic sampling was conducted within the district [17]. Details that could identify the unique personal information of the survey participants were not included. This study was approved by the Bioethics Review Committee of the affiliated institution. The inclusion criteria of the participants in this study were: ⅰ) “wage worker;” ⅱ) those with work type of “shift work;” ⅲ) those who had at least one-night shift per month; and ⅳ) those who were under the age of 60. The age limit of under 60 was applied to exclude the influence of the aging process on sleep disturbance, such as changes in melatonin secretion and circadian cycle imbalance [18]. A total of 1,678 participants who met the inclusion criteria were included in this study.
Variables
Sleep disturbance were assessed using the following three criteria: “Within the last 12 months, how often did you experience problems with: ⅰ) having difficulties in falling asleep; ⅱ) repeatedly waking up during sleep; ⅲ) waking up feeling tired or exhausted?” According to the criteria of the Diagnostic and Statistical Manual of Mental Disorders-5 (DSM-5) by the American Psychological Association (APA)[19], the responses “every day” and “several times a week” were coded as “sleep disturbance present,” whereas the responses “several times a month,” “rarely,” and “not at all” were coded as “sleep disturbance not present.”
The characteristics of working time, “shift type,” “extended work,” and “frequency of night work” were assessed using a questionnaire. First, according to the responses to the shift type question “Which of the following best represents the type of your shift work?,” the shift types were coded into split shifts, shift/rotation, and permanent shifts. Split shifts refer to a work schedule that divides a working hour into two or more working periods separated by periods of brake time, which require a shift interval of at least four hours per workday. For example, an employee works their first shift from 04:00 to 08:00, takes a break until 13:00, and then works their second shift from 13:00 to 17:00. Rotating shifts refer to work schedules that shift alternately between day, evening, and night work according to a predetermined rotating schedule. Second, extended work was assessed by asking, “How many days per month do you work more than 10 hours a day?” If participants recorded the number of days, they were coded as an extended work group; if not, they were coded as a nonextended work group. Third, the frequency of night work was measured by the number of days each participant responded to the question, “How many night shifts (working at least two hours between 10 pm and 5 am) do you work per month?”
Work characteristics, physical work demands, workload, emotional work demands, social support, and communication were assessed using questionnaires. Both physical work demands and workload were assessed with five items and four items respectively, on a 7-point Likert scale ranging from “throughout working hours” to “not exposed at all”. An example item for physical work demands is, “Does your work involve tiring or painful postures?” and for workload, “How often does your work require working at fast speed?.” The total scores were calculated by summing up each response to the questions and converting them to a score of 0–100; a higher total score indicated a higher physical work demand and workload. Cronbach’s alpha coefficient of the scales ranged from .75 to .80 [20] and were .64 in this study. Emotional work demands were measured with four questions: “Does your work involve complicated tasks?”, “Does your work deal with angry customers or patients?”, “Does your work place you into being emotionally unstable?” and “I need to hide my emotion in my working place.” The responses on dichotomous, or 5- or 7-point Likert scales for each question were converted to a score of 0-100 after summing the scores; with higher scores indicating higher levels of emotional work demands. Cronbach’s alpha coefficient for the scales was .71 [20] and was .50 in this study.
Social support was measured using three items: “My colleagues help and support me.”, “My boss helps and supports me.”, and “We are fairly treated in the workplace.” The responses on a 5-point Likert scale ranging from “always” to “not at all” were converted to a score of 0-100 after summing the scores. Cronbach’s alpha coefficient of the scales ranged from .76 to .85 [20] and was .66 in this study. Communication was measured with six items on a 5-point Likert scale (from “strongly agree” to “strongly disagree”). Example items are “Employees are recognized and complimented for good work in the workplace.” and “My colleagues and I are in good cooperation.” Responses were summed and converted to a score of 0-100, with higher scores indicating higher levels of social support. Cronbach’s alpha coefficient of the scales ranged from .68 to .87 [20] and was .86 in this study. General characteristics included gender, age, level of education, and monthly income.
Statistical analysis
As the work environment survey was based on a complex sample design, weighting was applied when analyzing the data to accurately estimate the population distribution [17]. Differences in sleep disturbance based on general characteristics, as well as work time and work characteristics, were analyzed using an independent t-test or χ2 test. No systematic pattern was observed for missing values of the variables, and multiple imputations were conducted for missing values before applying the inferential statistics. Multiple logistic regression analysis was conducted to identify the relationship between sleep disturbance and the work environment, with adjustment for covariates (demographic characteristics). The odds ratio (OR) and 95% confidence interval were calculated, and the fit of the model was confirmed using the Hosmer and Lemeshow test. We used a p-value of < 0.05 to determine statistical significance. SPSS/WIN 25.0 software was used for data analysis.