Working conditions survey
This study used a sample from the second wave of the Korean Working Conditions Survey (2010) conducted by the Korea Occupational Safety and Health Agency. The methodology and survey questionnaire used for the second Korean Working Conditions Survey is similar to those used in the European Working Conditions Survey, and the second survey built on investigations begun in the first survey (2006).
Briefly, the goal of the survey was to gather comprehensive information on Korean working conditions to shed light the nature and types of changes affecting the workforce and the quality of work-life for employees. The specific objective of the survey was to develop social and occupational health indicators for the working environment. We used multi-stage, random sampling from the Enumeration Districts in the 2005 Population and Housing Census. The survey was carried out at a number of different sampling areas, determined using probability proportional to population size and to population density. The survey data were collected from a nationally representative sample of the economically active population 15 years and older, excluding retirees, unemployed, homemakers, and students [17].
In this study, we defined the subjects as only ‘employed workers’, so we exclude ‘a self-employed worker’ or ‘an unpaid worker for familial business. With these exclusion criteria, among 10,019 survey interviews conducted at workers' homes between June and October 2010, the data from only 5,995 employed workers were used in this study.
The Institutional Review Board of Inha University Hospital approved the study protocol.
WHO five well-being index
Well-being was evaluated through the WHO Five Well-Being Index (the 1998 version). Although the index was originally designed to measure well-being in diabetic patients [18], its effectiveness has been supported in diagnostic depression screening [4] and evaluation of emotional well-being in patients with chronic diseases including cardiovascular diseases [18] and Parkinson’s disease [19], and in young children [20], and elderly adults [21].
The index consists of five positively worded items, each of which reflects the presence or absence of well-being. Subjects respond to questions about their positive feelings within the last two weeks on 6-point scale (0–5). A raw score lower than 13 out of 30 or an individual item score of 0 or 1 on any of the five items implies poor well-being [22], and a raw point score considerably below 13 may necessitate screening for depression with the Major Depression Inventory (under ICD-10) [4]. This study has evaluated the states of well-being of the subjects by classifying subjects with a raw score below 13 or a score of 0 or 1 on one of more items into the “poor well-being” group. Meanwhile, those who responded to all the items with a score of 2 or higher and had a total score higher than 13 were assigned to the “fair well-being” group [22].
Definition of independent variables
The survey collected information on several sociodemographic factors. Age, educational levels and monthly income were classified. Based on survey responses, the degrees of difficulty balancing income and expenses with total monthly earnings were divided into “difficult,” “somewhat difficult,” “somewhat easy,” and “easy.” Overall subjective health status, smoking status and alcohol assumption scale were also remarked. Subjects were grouped into “non-drinkers,” “moderate drinkers,” and “excessive drinkers” (i.e., individuals who consumed more than 2 glasses of alcohol a day).
Information on working condition factors was also collected and organized. Occupations were divided into 9 categories based on the Korean employment classification of occupation. Length of employment was divided into “<1 year,” “1–5 years,” “5–10 years,” “10+ years.” Employment types were divided into “contingent” and “regular” employees, and employment stability (“stable” or “unstable”) was determined by researchers based on the subjects’ responses. The presence and the absence of shifts were noted, and total weekly work hours were categorized into “<40 hours,” “40 hours,” ”40–60 hours,” and “60+ hours.” Based on the calculated difference between the actual and desired working time, we had categories of individuals who “lacked work time,” worked “10+ hours longer” than desired, worked “<10 hours longer” than desired or experienced “no difference” between actual and desired work time. Monthly work on the weekends was divided into “4+ days a month,” “1–4 days,” and “none.” Work condition satisfaction was divided into “satisfied” and “unsatisfied.” Degree of work stress was divided into the categories of “mild,” “moderate,” and “severe.”
Statistical analysis
All data were analyzed with the SPSS (version 14.0) after encoding was completed. A descriptive analysis was carried out on sociodemographic and working condition factors. Pearson’s chi-squared analysis was carried out between each factor and well-being. Logistic regressions yielded crude and adjusted odds ratios of working condition factors and the well-being. For logistic regression analysis, three models were used for adjusting effect of other factors. The model 1 was adjusted for sociodemographic factors (gender, age, education, monthly income, balancing income and expenses, health status, smoking, and drinking); the model 2 was adjusted for working conditional factors (type of occupation, number of years employed, employment type, employment stability, weekly working time, difference between actual and desired working time, monthly weekend work, satisfaction with working conditions, and work stress); and the model 3 was adjusted for both sociodemographic and working conditional factors. Pearson correlation analysis was used to test for multicollinearity among individual factors. The significance threshold was 0.05.
