Observational methods for the assessment of biomechanical risk factors: validity and limitations

S. Mattioli,¹ S. Curti² | ¹Department of Environmental Sciences and Prevention, University of Ferrara; ²Department of Medical and Surgical Sciences, University of Bologna, Italy

Introduction. Several observational methods have been reported in the literature, also applied in developing countries, to assess workers’ exposure to biomechanical overload. In the systematic review by Takala et al., published in 2010, the authors reported the existence of 30 observational methods.¹ However, this review did not consider measurement properties, risk of bias, and the intercultural adaptation process. Observational methods are numerous, and their number increases with the experience accumulated over the years of their use. Furthermore, the application of new and old methods should be based on their reliability and their ability to correctly identify not situations of absence or excess risk, but rather those in which it can be stated that, although there is exposure, only a small proportion of subjects might suffer a health problem in some way related to the exposure itself.

Objectives. The purpose of this work is to update Takala et al.’s 2010 systematic review of published studies on the validation of observational methods for assessing biomechanical risk factors.¹

Methods. The research conducted by Takala et al. at the time was replicated on PubMed/MEDLINE, following the general descriptions provided in the review published in 2010. The years following those already examined by the same author (i.e., from September 2008 to June 2025) were therefore explored. In the search strings,  terms related to biomechanical overload were used, such as workload, lifting, manual material handling, posture, ergonomic, biomechanical exposure, along with terms related to the musculoskeletal system and terms concerning evaluation methods, such as observational method, risk assessment, and task analysis. Finally, a filter was applied to identify systematic reviews published on the topic after that of Takala et al.

Results. Based on the search strategy developed for PubMed/MEDLINE, 612 citations were identified from September 2008 to June 2025. Among these, 20 literature reviews were identified, including two recent systematic reviews. The review by Mazloumi et al., published in 2025, evaluates only observational methods applicable to agriculture,² whereas the review by Valentim et al., published in 2024 and inspired by Takala’s work, explores the entire range of observational methods.³ Takala et al. reported 15 years ago that only two methods (Strain Index and ACGIH HAL) had validated predictive power through longitudinal studies, and also good or moderate intra- and inter-observer repeatability.¹ The recent review by Valentim et al. identified 23 additional methods compared to the 30 already identified by Takala et al. Moreover, the authors assessed the quality and risk of bias using the GRADE method (modified ad hoc). According to the authors, ALLA, HAL, OFFERA, and PATH (in both simplified and extended forms) are the methods for which there is the highest quality evidence regarding their reliability. Furthermore, PATH (extended form) and QEC are comprehensive in terms of both the criteria used and the body parts considered and are therefore suitable for evaluating many types of work tasks. According to the authors, EAWS is suitable for evaluating posture, force, and repetitiveness, while REBA and ROSA have proven measurement properties.

Conclusions. The systematic review by Valentim et al. reaffirms, as already reported by Takala et al., that to determine the method to be applied, it is necessary to consider the type of work activity being examined and the limitations imposed, including time and available resources. Highlighting the validity and reliability characteristics of different methods, Valentim et al. still suggest to use more than one method, which allows for comparisons between results. This will improve the completeness of the assessment, as already emphasized by Sala et al.’s 20 years of observations.⁴