Text

Introduction: The temporal structure of exposure, i.e., its variation (Mathiassen, 2006), plays an important role in work-related musculoskeletal disorders. While traditional metrics often fail to capture the temporal dimension, Exposure Variation Analysis (EVA) is specifically designed to account for both the intensity and the sequence duration of exposures. Since its introduction in 1991, different expressions of EVA have been used for several exposures. This is the first review to summarize its applications, methodologies, and future directions. We aimed to summarize and critically appraise the use of EVA to date, examine methodological variability across studies, and identify opportunities for future development. Thus, we conducted a literature search for studies published between 1991 and 2024 that explicitly utilized EVA. Studies were identified through citation tracking and database searches. Key information was extracted on exposure types, exposure and sequence duration intervals, EVA derivates, and statistical analysis approaches applied to EVA data.

Results: A total of 94 studies – primarily from occupational health, public health, and sports sciences – were included. Most were field studies (n = 54), followed by laboratory-based (24), combined field/laboratory (12), overviews (3), and simulations (1). EVA has most frequently been used for analyzing electromyography and postural data, and recently for physical behaviors, while examples exist for sports-related metrics (e.g., power output). We found substantial variability in how studies have defined categories for exposure intensity and sequence duration. Several studies simplified or post-processed the EVA matrix to extract summary measures, such as marginal distributions, variability indices, and centroid location. Only one study employed Compositional Data Analysis (CoDA) to appropriately handle the constrained structure of EVA matrices. Statistical models ranged from t-tests and analysis of variance to multivariate analyses.

Discussion: Although EVA has gained increasing attention across disciplines, its application remains inconsistent and methodologically fragmented. The large variety of exposure and sequence duration intervals limits comparisons between studies and highlights the need for standardization and further development. Furthermore, many studies that use conceptually similar approaches—such as bout analysis—do not reference EVA, limiting the visibility and advancement of the method.

Conclusions: EVA is a versatile tool for quantifying changes in exposure over time and has been applied across several disciplines. However, it has primarily been used to document exposures rather than analyzing associations with health outcomes. Future research should focus on standardizing interval sets for different exposure types, developing and validating EVA derivates, analyzing long-term health effects of different EVA structures, and developing appropriate statistical models that reflect these analytical needs. These efforts are essential for enabling consistent, interpretable, and health-relevant applications of EVA in occupational and public health research.