Objectives: This study compared the cost-efficiency of observation and inclinometer assessment of trunk and upper arm inclination in a population of flight baggage handlers, as an illustration of a general procedure for addressing the trade-off between resource consumption and statistical performance in occupational epidemiology.

Methods:  Trunk and upper arm inclination with respect to the line of gravity were assessed for 3 days on each of 27 airport baggage handlers using simultaneous recordings by inclinometers and video.  Labour and equipment costs associated with data collection and data processing were tracked throughout.  Statistical performance, in terms of the inverse of the standard deviation and root mean squared error of the group mean exposure, was computed from the variance components within and between workers, and bias (with inclinometer assumed to produce ‘correct’ inclination angles).  The behavior of the trade-off between cost and efficiency with changed sample size, as well as with changed logistics for data collection and processing, was investigated using simulations.

Results:  At similar total costs, time spent at trunk and arm inclination angles greater than 60 degrees as well as 90^th percentile arm inclination were estimated at higher precision using inclinometers, while median inclination and 90^th percentile trunk inclination was determined more precisely using observation.  This hierarchy persisted in a scenario where the study was immediately reproduced in another population, while inclinometry was more cost-efficient than observation for all three posture variables in a scenario where data were already collected and only needed to be processed. Observations showed to be biased relative to the –assumed to be correct – inclinometer data, and so inclinometry became the most cost-efficient option for all posture variables and irrespective of scenario when statistical performance was measured by bias and precision combined.

Conclusions: Inclinometers were more cost-efficient in use than observation for two out of three posture metrics investigated when statistical performance was measured only in terms of precision. Since observations were biased, inclinometers consistently outperformed observation when both bias and precision were included in statistical performance. The presented general model for assessing cost-efficiency may be used for designing exposure assessment strategies with considerations not only to statistical criteria, but even to costs. The empirical data provide a specific basis for planning assessments of working postures in occupational groups.