Crosswalking Disciplines: Epidemiology Study Designs for Occupational Health Psychology

In Occupational Health Psychology (OHP), researchers often seek to understand the causal relationships between workplace factors and employee well-being. As a graduate student in epidemiology, I see great potential for epidemiology methods in OHP. The epidemiology discipline also has a strong focus on causal inference, where discussion on proper methodology, uncertainty, and validity comes foremost. Much of the theory behind causal inference began with Bradford Hill’s 9 “Considerations for Causation”, with the most relevant being temporality, of having a relationship with time (Hill, 1965). For more modern theory behind causal inference in epidemiology, readers are encouraged to seek out Rubin’s potential outcomes model theory (Rubin, 2005)

Ployhart and Vandenberg (2010) have summarized the absolute need for an increase in rigorous longitudinal research. They point out the issues relating to the theory, design, and analysis of change, and even offer helpful guidance on study design in the form of checklists (Ployhart & Vandenberg, 2010). This call for studying change echoes the need for temporal research that seeks to establish causality between workplace stressors and strains. Fortunately, there has been a marked increase in longitudinal studies in OHP. However, improvements can be made to longitudinal study design and application. Taris and Kompier (2014) recognized this gap and called for OHP to conduct more longitudinal studies.

One example of a longitudinal study would be the long-term Occupational and Environmental Health Cohort Study, where various occupational stressors and impacts on health are examined in a population-based cohort with varying follow-up times (Slottje et al., 2014). However, long-term cohort study designs are not easy to pull off. It is important to consider alternative study designs to the traditional cohort design given the unique research questions found in the field of OHP, which is why I offer examples of alternative epidemiological designs and show how they are relevant and helpful to OHP.

Epidemiology Study Designs

Natural experimental designs are quasi-experimental designs that take advantage of a naturally occurring event. Unlike randomized controlled trials where researchers randomly assign participants to different conditions, natural experiments leverage circumstances such as policy changes or natural disasters to create comparable groups for analysis. This allows researchers to perform observational studies without the same ethical or logistical challenges of conducting a randomized experiment (Craig et al., 2017). Another advantage of the natural experiment is that it can occur in many different study settings, whether an intervention is being tested or a longitudinal cohort study is collecting data. However, natural experimental designs are only useful when a study is already ongoing or there is pre-existing data available.

A great example of this in OHP is when Moen and colleagues were evaluating the effectiveness of an intervention to increase employee’s flexibility and support. Approximately halfway through their study, a company merger event occurred, and they leveraged the naturally occurring event to observe the effects of their intervention before and after the merger. This study provided valuable and unique insights, showing that the intervention had a limited effect for those who received it after the merger (Moen et al., 2016).

Case-crossover studies have great potential in OHP. Contrasting with a case-control design clarifies the case-crossover design. An example of a case-control study would be a study where those with a strain are treated as cases and their stressor history is compared to individuals without that strain (i.e., the controls). This contrasts with a case-crossover study where an individual serves as both the case and control for themselves. Multiple measurements are taken routinely and when an acute event occurs the stressor measurements leading up to that event could be compared to another time frame for that same individual (Maclure, 1991). This design could be highly practical in OHP studies that use daily diaries.

This design has been successfully used in OHP before in a Swedish study that investigated  work-related psychosocial events as triggers for sick leave (Hultin et al., 2011). Investigators had a cohort of 1430 employees and collected baseline data, daily reports from the workplaces on start and end dates of all sick-leave periods during follow-up (varying 3 to 12 months between different workplaces), and telephone interview data. The authors found that the risk of taking sick leave was increased when individuals had recently, within the prior two workdays, been exposed to problems in their relationship with a superior.

Nested case-control studies are nested within a cohort, comparing individuals who develop a particular outcome/strain with those who do not develop that outcome but are in the same cohort. These studies are efficient for examining rare stressors or outcomes. This design can be particularly feasible when there is already an available cohort study underway, allowing there to be relatively few financial and logistical challenges (Ernster, 1994).

The nested case-control design has been used previously in OHP with sawmill workers in Western Canada who were enrolled in a cohort study. Investigators found that workers with a shorter duration of employment (i.e., organizational tenure) had increased odds for acute reactions to stressors (Ostry et al., 2006).

Choosing a Research Design

Each of these epidemiology study designs offers unique strengths and limitations that OHP researchers should consider.

• Cohort studies provide strong evidence for causality but require immense resources.
• Natural experiments offer real-world insights but are not always an option without a natural event occurring.
• Case-crossover studies are efficient for studying acute effects but rely upon strong assumptions about transient exposure effects and are most appropriate when studying acute outcomes.
• Nested case-control studies are feasible but require an existing cohort.

The choice of which design is the right one for an OHP study depends on the research question which can be best informed by careful consideration of the variables at play and available resources.

Conclusion

Incorporating epidemiology study designs into OHP research enhances the ability to infer causal relationships between workplace stressors and employee well-being. By using innovative study designs, researchers can deepen their understanding of occupational health outcomes and inform evidence-based interventions. This blog post serves as a call for OHP researchers to continue going beyond cross-sectional designs and offers practical ways to use epidemiology designs to improve scientific rigor in OHP.

*Author Note: I am deeply appreciative of Dr. Kimberly French’s expert support in the writing of this work and her strong encouragement to seek out a means to contribute to the field.