Help-seeking for mental health problems by employees in the Australian Mining Industry

This cross-sectional study used a quota sampling method with the aim of recruiting a representative sample of employees across the Australian coalmining industry. The approach adopted a number of stratification variables to ensure coverage of coal mining locations (New South Wales [NSW] and Queensland [QLD]; the two primary coal producing states in Australia), mine type (underground or open-cut), and the two primary types of employee commute arrangements (daily commute or long distance commute [DIDO/FIFO]).

Mining companies were approached to provide consent at the company level, which primarily occurred through contact with occupational health and safety managers, with all companies contacted consenting to participate. After company consent, the research team liaised with managers from individual mine sites, and provided an overview of the project to mine site management and occupational health and safety staff, and requested consent at the mine management level. Once consent was received from management, the research team met with key delegates from each mine site to organize the logistics of data collection.

The data collection protocol was designed to minimize disruption to production, and to accommodate unique logistical considerations of each site. Where possible, the data collection occurred as a component of the sites’ routine training days, which are rostered days assigned for professional development. At sites where training days were unavailable or infrequent, data collection occurred while participants were on shift or during their daily pre-shift meetings. All accessible staff onsite during the specific site visit for data collection were invited to participate.

Two weeks prior to data collection, each participating site was sent a set of study information materials (e.g. PowerPoint slides, posters/flyers) to be displayed in employee common areas to promote awareness of the project. The data collection process involved a research team member visiting the site, providing a brief presentation that gave an overview of the research, as well as a written information statement outlining the research purpose. This material stated that the research was voluntary, confidential, and that participants were free to withdraw at any time. Participation involved completing a brief paper-based survey that took an average of 15 min to complete; return of a completed survey was considered implied consent. Participants were asked to provide a self-generated code (consisting of initials and day/month of birth), to allow potential cross-linkage with future surveys. While it was not anticipated that the content of the survey would cause distress, all participants were provided with a pocket size information card that contained the contact details of a number of different free-to-access support services.

The survey was conducted between December 2013 and March 2015.

The survey instrument used for data collection is provided in Additional file 1.

To determine factors associated with help seeking, a series of conceptually related variables were grouped into three separate categories (socio-demographic; predicted need for mental health services; and workplace factors and attitudes). This categorization reflected a theoretical interest in determining the independent effect of workplace factors and attitudes on the likelihood of contacting professional or non-professional sources of support, after controlling for the participant’s socio-demographic characteristics and predicted need for professional mental health services.

Socio-demographic information collected included the participant’s age, gender, relationship/marital status, highest level of education and whether the participant had any dependent children. The number of, and frequency of contact with social connections, and formal and informal group membership was also determined using the Social Network Index [18], with participants scored on a 4-point scale (low, medium, medium-high, high).

To derive an estimate of the need for professional mental health services, the current study used an adapted form of the Predicted Service Need Index (PSNI) [7, 19]. The PSNI takes into account a combination of the participant’s current symptomatology, potential effects from chronic health conditions, and current unhealthy behaviors, which may all reflect an increased need for professional mental health services. Specifically, the PSNI is an aggregate score of participants responses across five health status measures, including: the Kessler-10 (K-10) [20]: a measure of current psychological distress; their self-perceived overall rating of physical and mental health: both assessed using a 5-point Likert scale ranging from ‘poor’ to ‘excellent’; the Alcohol Use Disorders Identification Test (AUDIT) [21]: a measure of problematic alcohol use; and their current smoking status: a single item that measured whether the participant currently smoked cigarettes daily. Each participant received a PSNI score by adding previously determined integer weights (see Perkins et al. [19]) for each of the five health status measures, giving a possible PSNI score range from 0 to 9, with higher scores indicating an increased predicted need for professional mental health services. In the current study, PSNI scores were used both as a dimensional measure (i.e., simple index of increasing predicted need) and a categorical measure, using three previously described [7, 19] categories of need for professional services: low (0–1); medium (2–5); and high (6–9).

The majority of questions regarding workplace factors were measured using single-item questions including: years working in the mining industry, mine type, commute type, employment category, employee type, shift type, and shift length. The proportion of time at work was also measured based on the participant’s most typical roster, and was calculated by dividing the number of consecutive days at work by the sum of the consecutive days at home plus the number of consecutive days at work.

The survey also included a series of questions to measure participants’ attitudes towards working in the mining industry. Preliminary principal components analyses were conducted to guide the derivation of subscale scores for the following measures of participants’ attitudes:

Workplace attitudes were also assessed using two single-item questions that measured the concern participants had about losing their job and the perception of stigma in the workplace. Stigma was measured by participants self-reporting whether they felt an employee experiencing mental illness would be ‘treated poorly in the workplace if people found out about it’. Perceived stigma was measured on a five-point scale ranging from ‘strongly disagree’ to ‘strongly agree’, with responses categorized as: low stigma (strongly disagree or disagree); unsure (unsure); or high stigma (agree or strongly agree).

Data were analysed using conventional statistical packages, Microsoft Excel (Version 14) and the Statistical Package for Social Sciences (IBM SPSS version 22; Armonk, NY, USA). Descriptive analysis was used to characterise the socio-demographic and health status of participants. The initial analysis also included a description of contact patterns (both professional and non-professional) associated with discussions about participants’ own mental health within the last 12 months, including the proportion of people who contacted each source of support and the frequency of contact. To facilitate discussion, the types of contact are ranked in order of mean PSNI scores.

The derivation of subscale scores for several of the workplace attitude questions (i.e. satisfaction with work; perceived commitment of mine to the mental health of employees; and the primary reasons for working in mining) was guided by preliminary principal components analyses, based on sets of questions contributing to factors with an eigenvalue above 1.00.

To explore the factors associated with professional and non-professional contacts within the preceding 12 months, we used two hierarchical logistic regression models, with a predetermined order of entry for all associative variables. To examine the unadjusted contributions, in model one, all socio-demographic variables were simultaneously entered at step one, followed by each of the workplace and employment characteristics, which were added separately at step two. For model two, all socio-demographic variables were simultaneously entered at step one, followed by the PSNI category at step 2, and all workplace and employment characteristics added simultaneously at step 3. For all analyses, we report Adjusted Odds Ratios (AOR) with 99 % confidence intervals (CI).