Data
The cross-sectional study used data from the 2015 National Health Interview Survey (NHIS), a nationally representative household survey targeting the non-institutionalized, civilian US population. The NHIS uses multistage random sampling accounting for clustering, stratification, and weights to achieve the national representation [28]. The NHIS 2015 database was the most recent data that contains the key outcome variable in the study, workplace harassment or bullying. As the current study was focused on mental health issues related to the workplace, the study population was limited to working adults ages 18 or above. A total of 19,456 individuals meeting the inclusion criteria were included in the analysis.
Measures
Outcome variable
The key outcome was a dichotomized measure of SPD using the Kessler Psychological Distress Scale (K6). The K6 was a widely used tool for identifying severe psychological distress among general populations and was well-validated, consistent with the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) to detect DSM-IV diagnoses [29-30]. The K6 consists of six questions capturing psychological symptoms during the past 30 days. The questions are: 1) how often did you feel - so sad that nothing could cheer you up? 2) how often did you feel - nervous? 3) how often did you feel - restless or fidgety? 4) how often did you feel - hopeless? 5) how often did you feel - everything was an effort? 6) how often did you feel - worthless? Each question has five-level scores from 0 to 4, leading to a total score ranging from 0 to 24. The K6 scores equal or above the cut-point of 13 indicate SPD, representing a diagnosis of a serious mental health condition [30]. Then, the scores were dichotomized as having SPD or not.
Covariates
Key covariate
The key independent variable for this analysis was a binary indicator of workplace bullying or harassment. The variable was based on a dichotomous response to the survey question “During the past 12 months, were you threatened, bullied, or harassed by anyone while you were on the job?”
Additional covariates
Job-related characteristics were included based on the unified model [25]. Participants were asked, “Please tell me whether you strongly agree, agree, disagree, or strongly disagree with this statement? – I have enough time to get the job done”. The four responses were dichotomized for having enough time to get a job done or not. Participants further answered four-level scale questions: “The demands of my job interfere with my personal or family life”, “My job allows me to make a lot of decisions on my own”, and “I can count on my supervisor or manage for support when I need it”. Job interference with personal or family life was dichotomized. Job support was an indicator variable for having needed support in the workplace, and job control represented workers’ engagement in job-related decisions. Worktime shift had four categories of daytime, evening, night, and rotating in response to the question, “Which of the following best describes the hours you usually work?”. Additionally, hourly paid was included given it is a potential risk factor for workplace bullying [31].
The demographic characteristics included age, gender, race/ethnicity, marital status, income, and health insurance. Age was categorized as 18-29, 30-44, 45-59, and 60 or above. Race/ethnicity had four categories, such as white, black, Hispanic, and minority. Marital status was a categorical variable reflecting married, widowed/divorced/separated, unmarried couple, and unmarried. Income had three-level categories: <$49,000, $50,000-$99,000, and >=100,000. Health insurance was dichotomized, insured or uninsured. Also, participants’ health behaviors and health conditions were considered. Both smoking and drinking status were categorized into never, former, and current, and physical activity was dichotomized for sufficient physical activity referring to the recommendation by the Healthy Americans [32]. The study also incorporated a series of chronic conditions in analysis, such as coronary heart disease (CHD), heart attack, chronic obstructive pulmonary disease (COPD), asthma, cancer, and diabetes.
Statistical analysis
The characteristics of the study population and bivariate differences in SPD were examined using Pearson’s Chi-Square test. In regression analyses, unadjusted odds ratios were estimated from logistic regression models and used for examining the association between SPD and workplace bullying or harassment. We subsequently fitted multivariable logistic regression models, controlling for covariates, to estimate adjusted odds ratios for SPD. Additional analysis was performed to obtain the adjusted probabilities for SPD by workplace bullying or harassment status, and estimated figures were displayed graphically. Lastly, we calculated stratified odds ratios stratified by age, gender, and race/ethnicity using interaction effects between SPD and those three key population characteristics, while controlling for all other covariates. Data management and subsequent analyses were conducted using the Stata software, version 15.