Elsevier

Applied Ergonomics

Volume 47, March 2015, Pages 242-252
Applied Ergonomics

Is what you see what you get? Standard inclinometry of set upper arm elevation angles

https://doi.org/10.1016/j.apergo.2014.08.014Get rights and content

Highlights

  • Meticulously set upper arm elevation angles were measured by standard inclinometry.

  • A downward bias was found in the inclinometer results, particularly at angles > 60°.

  • A 2-point, subject-specific, linear calibration was effective at removing bias.

  • Calibration equations were developed for studies lacking 2-point calibration data.

  • Elevations gauged by inclinometry and observation need adjustment to a common scale.

Abstract

Previous research suggests inclinometers (INC) underestimate upper arm elevation. This study was designed to quantify possible bias in occupationally relevant postures, and test whether INC performance could be improved using calibration.

Participants were meticulously positioned in set arm flexion and abduction angles between 0° and 150°. Different subject-specific and group-level regression models comprising linear and quadratic components describing the relationship between set and INC-registered elevation were developed using subsets of data, and validated using additional data.

INC measured arm elevation showed a downward bias, particularly above 60°. INC data adjusted using the regression models were superior to unadjusted data; a subject-specific, two-point calibration based on measurements at 0° and 90° gave results closest to the ‘true’ set angles.

Thus, inclinometer measured arm elevation data required calibration to arrive at ‘true’ elevation angles. Calibration to a common measurement scale should be considered when comparing arm elevation data collected using different methods.

Introduction

It is generally accepted that estimates of working postures become increasingly more correct as one moves from subjective assessment to observational methods to direct measurement tools (van der Beek and Frings-Dresen, 1998, Winkel and Mathiassen, 1994). Thus, a measurement hierarchy has emerged in which data from higher echelon measurement tools are inherently considered to be ‘better’, that is, more precise (repeatable) and less biased (closer to the actual value) than data from measurement tools at lower levels. From a precision standpoint, this hierarchy is well supported: tools at lower levels are often associated with a larger methodological variability than those at higher levels. For instance, considerable variability has been reported within and between observers rating postures from the same video recordings (Mathiassen et al., 2013, Rezagholi et al., 2012), whereas direct measurement tools, for example, inclinometers, have excellent reported repeatability (Hansson et al., 2001, Hansson et al., 2006). However, the ability of a device to produce a faithful measurement depends on the conditions and contexts in which the measurements are made, and not simply on the technical attributes of the measurement device itself (Portney and Watkins, 2009). Thus, if a device interacts with the system it is intended to measure, measurements may be biased. For example, when using an inclinometer to measure joint angles, measurement bias may be introduced if the inclinometer moves relative to the underlying skeleton. By extension, careful consideration is required regarding whether exposure estimated using different methods are directly comparable; for example, postures assessed using inclinometers and observation. If inclinometer measurements do not agree with estimates of the same joint angle obtained using standard practice observation, it must be considered how to transform measured values into a common scale – ideally, one which comes as close as possible to ‘true’ values. If disagreement is considerable, comparison between studies based on observations and inclinometry can be severely compromised, and thus the need for a bias-corrected common scale is vital.

In 1937 (translated to English in 1972), Goldmeier demonstrated that subjects could always correctly visually identify a 90° angle when presented with simple drawings of 87°, 90°, and 93° angles provided that the drawings were presented in a normal orientation. This excellent ability of humans to correctly identify normal right angles has been dubbed the ‘Goldmeier effect’ (Ferrante et al., 1995). Applying the Goldmeier effect to the human body, it would be anticipated that, if a person were asked to identify a 90° shoulder abduction angle, the selected elevation angle would look approximately like that shown on the left side of Fig. 1. Similarly, a 90° shoulder flexion angle is expected to look approximately like the angle shown on the right side of Fig. 1. This expected consistency between observers suggests inherent construct validity to observation, at least at 90° angles: arguably, this could therefore be considered a ‘true’ 90° angle. By extension, this would mean that direct technical measurement methods should also assess this angle to be 90°. Findings published by Genaidy et al. (1993) showing that observers were fairly proficient and almost unbiased (throughout the 0–180° range of shoulder flexion) at correctly judging ‘true angles’ from still frames taken from video shot perpendicularly to the worker support this notion.

Over the last decade, tri-axial accelerometers have frequently been used as inclinometers to assess upper arm elevation angles with respect to the line of gravity (for example: Bernmark and Wiktorin, 2002, Delisle et al., 2006, Fethke et al., 2011, Hansson et al., 2010, Korshøj et al., 2014, Leijon et al., 2005, Wahlström et al., 2010, Veiersted et al., 2008). Inclinometers are attractive direct measurement devices due to their relatively low cost (Trask et al., 2013), ease of use, and highly portable characteristics.

A 2002 paper by Bernmark and Wiktorin investigated the validity of utilising tri-axial accelerometers as inclinometers (INCs) for the purpose of measure arm postures and movements. To achieve this goal, INC angle measurements were compared to angle measurements recorded from an optoelectronic (OPT) measuring system at fixed arm elevation angles of 0, 45, 90, 135 and 180° that were determined by meticulous, but unassisted, observation by a single researcher (personal communication – Eva Bernmark). The paper presented a figure showing a strong correlation between the INC and OPT measured arm elevation angles (Bernmark and Wiktorin, 2002). The figure also showed that both systems underestimated the expected inclination angles at, and above, 90° arm elevation. The authors have generously shared the data behind their original figure, and we re-plotted the INC measurement data with respect to the expected angle data to highlight this additional finding (Fig. 2). Non-published data from within our research team have also suggested a similar trend of underestimated arm elevation angles across multiple brands of INC systems and INC mounting protocols. This underestimation occurs despite the excellent accuracy of tri-axial accelerometer INC systems during static testing in a rigid rack; for example, the Logger Teknologi AB system (Logger Teknologi HG, Åkarp, Sweden) has a reported ‘accuracy’ of 1.3° and a ‘reproducibility’ of 0.2° during static testing (Hansson et al., 2001), and angles measured using the Virtual Corset system (Microstrain Inc., Vermont, USA) were reported to deviate no more than 2° from correct values (Amasay et al., 2009).

The purpose of the current study was therefore to evaluate the extent to which arm elevation angles measured using standard tri-axial inclinometry systematically underestimate ‘true’ elevation angles, as determined from meticulous, assisted observation positioning of the arm in different postures, and whether a possible bias could be effectively adjusted for using regression models. Two different INC mounting locations on the upper arm were investigated to determine whether anatomical placement has an effect on estimation accuracy.

Section snippets

Subjects

A total of 19 participants (12 males, 7 females) were recruited by advertisement at three educational institutions in Gävle, Sweden (Table 1). Subjects were excluded if they had any ailment preventing them from comfortably moving through a full range of shoulder motion (arms at side to arms overhead) or to their maximum trunk flexion angle (trunk data not reported in the current study). The study was approved by the Regional Ethical Review Board in Uppsala, and all subjects signed an informed

Results

Data from the Cr-VM INC were successfully collected for all subjects and days. A technical issue with the data logger for the Ca-LT INC resulted in loss of data for 6 of the 19 subjects on day 1.

Data from one representative subject are given in Fig. 5, showing the relationships between the set angles, unadjusted INC measurements, and calibrated INC data for both arm abduction and arm flexion. Increased bias at higher elevation angles for the unadjusted INC data is clearly shown for the INCs in

Discussion

Our data demonstrate that arm elevation angles recorded using standard inclinometry systematically underestimate what we argue to be a ‘true’ assessment of arm elevation angle, i.e. by meticulous, assisted observation, particularly at elevation angles greater than 60°. By extension, our results question the notion that different measurement tools assess upper arm postures according to a common scale, and further, whether un-calibrated inclinometers provide the most accurate data possible. Our

Conclusion

A systematic underestimation of upper arm elevation measured using standard inclinometry was found, particularly at larger elevation angles. The bias could be adjusted for by applying regression calibration models; the most effective model found was a subject-specific, two-point linear approach that utilised the slope from the origin to the mean of inclinometer recorded values at 90° arm abduction and 90° arm flexion. This model outperformed quadratic models using group-level data as input to

References (34)

  • K.B. Veiersted et al.

    Effect of an intervention addressing working technique on the biomechanical load of the neck and shoulders among hairdressers

    Appl. Ergon.

    (2008)
  • S. Bao et al.

    Interrater reliability of posture observations

    Hum. Factors

    (2009)
  • A. Delisle et al.

    Comparison of three computer office workstations offering forearm support: impact on upper limb posture and muscle activation

    Ergonomics

    (2006)
  • A.M. Genaidy et al.

    Can visual perception be used to estimate body part angles?

    Ergonomics

    (1993)
  • E. Goldmeier

    Über Ähnlichkeit bei gesehenen Figuren

    Psychol. Forsch.

    (1937)
  • E. Goldmeier

    Similarity in visually perceived forms

    Psychol. Issues

    (1972)
  • G.A. Hansson et al.

    Validity and reliability of triaxial accelerometers for inclinometry in posture analysis

    Med. Biol. Eng. Comput.

    (2001)
  • Cited by (20)

    • Reducing postural load in order picking through a smart workwear system using real-time vibrotactile feedback

      2020, Applied Ergonomics
      Citation Excerpt :

      The threshold values for the upper arm elevation angles (i.e., 30° and 60°) were based on the action (threshold) values proposed by Arvidsson et al. (2017) for the prevention of WMSDs. The two IMUs were placed bilaterally in the t-shirt upper arm pockets with the upper edge of the IMUs approximately at the insertion of the deltoideus muscle (Jackson et al., 2015), while the trunk IMU was placed at the upper back at the level of the thoracic vertebrae 1–2 (Korshøj et al., 2014). The two actuation units were mounted with straps: the first on the dominant upper arm below the IMU, and the second approximately at the sternum (Fig. 2A).

    • Accuracy of angular displacements and velocities from inertial-based inclinometers

      2018, Applied Ergonomics
      Citation Excerpt :

      Finally, this study considers strictly sensor error. A recent study, for example, demonstrated that accelerometer-based inclinometers may underestimate inclination measurements under static conditions, particularly at angles >60° (Jackson et al., 2015). Differences in error magnitudes due to measurement methodology, such as differences in the local coordinate frame defined using anatomical landmarks in comparison to the sensor local coordinate frame and errors due to soft tissue artifacts, were not considered.

    • Validity of a small low-cost triaxial accelerometer with integrated logger for uncomplicated measurements of postures and movements of head, upper back and upper arms

      2016, Applied Ergonomics
      Citation Excerpt :

      This may be due to that the relative position of the inclinometer to the underlying bone, will change throughout the range of motion, to different extents at different arm elevations, i.e. soft tissue artifact. In a recent study, Jackson et al. showed an underestimation of about 10° at instructed arm elevations of 90° (Jackson et al., 2015). However, our experience shows a difference of <1° at instructed arm elevations of 90° (Hansson, 2015).

    View all citing articles on Scopus
    View full text